Prometheus. I sat with that for a moment. Normally, a new venture gets named after something relevant to the company’s activity, usually a mix of words tied to the product or industry. But in this case, Bezos chose a name pulled straight from Greek mythology, after a titan rather than a business theme.

In fact, he was the cheeky Titan (I had to look this up) who defied the gods by stealing fire from Mount Olympus and gifting it to humanity. And for his trouble, he got chained to a rock on eternal barbecue duty, courtesy of Zeus’s HR department.

Strip away the mythic flair, and Project Prometheus emerges as a calculated strike at the heart of an industry stuck in slow motion. A six point two billion dollar war chest, with Bezos fronting a significant slice, fuelling an AI venture that’s already poached nearly one hundred minds from the labs of Meta, OpenAI and Google DeepMind.

Something straight out of Meta’s recent playbook, where they went on a hiring spree to beef up their own AI ambitions.

So he’s back in play, and I didn’t think too much of it at first, apart from the usual “maybe he’s just bored with the Med, retail or rockets.” But the detail that shifted me from that to “okay, this is actually big news” was learning that his new venture is zeroed in on AI for the physical economy. Real tools that simulate and improve the gritty realities of engineering and manufacturing.

Think proprietary models that crunch physics data to virtually test rocket fuselages for thermal stress or check new EV battery chemistries for long term wear. All of it still speculation, but if it works, it could cut prototyping timelines from years to months and trim costs by fifty to seventy percent in the most expensive parts of R&D.

Now, this is all good and well, but where does this actually get applied? What’s the gameplay here?

He is betting big on what I would think is AI that whispers sweet nothings to physics itself, on a mission to drag American manufacturing back from its rusty exile in Shenzhen’s shadow.

Think proprietary models that crunch physics data to virtually test rocket fuselages for thermal stress or check new EV battery chemistries for long term wear. All of it is still speculation, but if it works, it could cut prototyping timelines from years to months and trim costs by fifty to seventy percent in the most expensive parts of R&D.

This is a big play and probably something you would expect from Elon Musk’s playbook, because it’s the meat and potatoes of forward thinking innovation. And I would be remiss if I didn’t admit that part of me wonders if Bezos was stepping out on an overambitious limb.

But then…

You see his operational co pilot, or officially Co-CEO and Co-Founder. It’s Vik Bajaj, a physicist chemist with proper moonshot pedigree. Co founder of Alphabet’s Verily, which is Google’s life sciences arm, and a key architect at Google X, where he helped spearhead both self driving bets and early biotech work.

So Bajaj’s brief is to helm the science while Bezos drives the scale, a setup that mirrors Amazon’s early playbook of relentless execution layered over audacious vision.

Now here’s where the pot gets stirred, because you can’t talk about this without touching the China question. If Prometheus actually works, automation becomes the great equaliser. Labour costs suddenly matter a lot less. Whether a welder in the U.S. charges thirty dollars an hour or someone in Shenzhen charges five… the machine doesn’t care.

Picture AI-run factories pumping out chips and hardware at the same pace as China’s best, boosted by CHIPS Act money and a few well placed Bezos-built plants in Ohio running twenty four seven with robot crews that never call in sick and don’t go on strike. That is the dream scenario.

But poke that bear a little harder. China is not going to sit there and watch. Huawei is already building out its HarmonyOS world, their state backed AI fabs are well into the hundreds of billions, and they control most of the global battery supply. Around sixty percent. The U.S. still trails badly in raw materials. Rare earths? China handles most of that. And you still have the data problem. Prometheus needs huge piles of physics data, and export controls could choke some of the input that these models depend on.

So the only way this works in the long run is if Bezos builds real moats. Maybe he keeps some of the best stuff inside Blue Origin. Maybe he pushes for tight supply chains that Washington will support. Without that, it starts feeling like David squaring up to a dragon that breathes fire.

Could he pull it off? Sure. If he manages to hand out this new “fire” without burning down the village. And if not… well, in true Titan fashion, we might all just end up warming our hands at the blaze.

GPT 5.1 landed with solid upgrades.
Kimi pushed out K2 and people jumped straight into testing the reasoning boost.
A few labs dropped new benchmarks.

But that is not the real headline.

The real headline is that Anthropic released the full report of a live cyber attack that ran in September.

The back story is that a state linked espionage team used Claude in a real intrusion.

Not theory.
Not a demo.
A real operation with real targets.

This is the first time we have seen a fully documented case of an AI system helping to carry out intrusion steps in the wild.

We have officially moved beyond the “what if” stage.
It is now clear that AI cyber attacks have begun.

Let’s break down what happened and what it means for platforms and engineers.

What GTG 1002 Actually Did

The report identifies the operators as GTG 1002, a state linked espionage team known for targeting high value environments.

Here’s how Anthropic described the intrusion in their official November 2025 report.

In mid-September 2025, we detected a highly sophisticated cyber espionage operation conducted by a Chinese state-sponsored group we’ve designated GTG-1002 that represents a fundamental shift in how advanced threat actors use AI. Our investigation revealed a well-resourced, professionally coordinated operation involving multiple simultaneous targeted intrusions. The operation targeted roughly 30 entities and our investigation validated a handful of successful intrusions.
Source: Anthropic Full Report, November 2025

Their targets included banks, tech firms, and government bodies.
Roughly thirty organisations in total.

These were not random targets.
These were chosen for access, insight, and strategic value.

This is now what automated cyber threat campaigns look like.

A small team suddenly has the power of a much larger threat operation.

The Architecture Behind the Operation

The architecture of this campaign tells you everything about how far threat actors have already gone with AI driven operations.

GTG 1002 did not run a simple script or a noisy mass scan.
They built a structured campaign with clear phases, long lived model sessions, and a split between model autonomy and human oversight.

Claude handled most of the operational work.
The human operators only stepped in at strategic points where judgment mattered.

This is how the report describes it:

“The campaign proceeded through structured phases, where AI autonomy increased progressively while human oversight remained concentrated at strategic decision gates.”
Source: Anthropic Full Report, November 2025

This means the operation had a real hierarchy.
Claude did the scanning, sorting, checking, mapping, testing, and summarising.
The humans chose targets, adjusted strategy, and decided what to do next.

Across the campaign, the pattern looked like this:

Claude ran two to six hour blocks of autonomous activity.
Humans stepped in every five to twenty minutes to approve the next move.
Then Claude ran the next sequence.

This architecture let a small team work at a speed that usually requires a large crew.
Much of the workflow ran like a distributed system.

Claude acted as the worker nodes, processing small tasks and producing structured output.

The human operators acted as the orchestrators, linking outputs into the bigger chain.

It is also clear that the attackers understood the model’s tool ecosystem.
They used browser tools, scanning utilities, payload testers, and structured reporting to keep the workflow cohesive.

Everything was broken into small tasks so the model would treat each step as normal.

Once you understand the architecture, the lifecycle makes sense.
The attack was not random.
It moved through phases.
It expanded autonomy with each step.

And it used human oversight only where strategic decisions were needed.

Anthropic’s Response

“Upon detecting this activity, we immediately launched an investigation to understand its scope and nature. Over the following ten days, as we mapped the severity and full extent of the operation, we banned accounts as they were identified, notified affected entities as appropriate, and coordinated with authorities as we gathered actionable intelligence.”
Source: Anthropic Full Report, November 2025

Anthropic’s response was fast and structured.

They picked up the unusual behaviour patterns early and moved into an investigation straight away.

The next ten days were spent mapping the scale of the operation, confirming what was real, and verifying how far the attackers had progressed.

As they traced the activity across accounts, they banned each one as soon as it was confirmed.

They notified the organisations that were affected.

And they worked with the relevant authorities while the intelligence was being collected.

This is an important part of the story.
The detection came from behaviour, not prompts.
And the response followed the same pattern you would expect from any security incident:
detect, verify, isolate, notify, and contain.

This also shows that the defenders moved at speed.
And speed matters now more than ever.

Claude’s Network Feature Flaw

The report surfaced another issue.

A separate flaw that involved Claude’s network features and how they could be manipulated through indirect prompt injection.

Security researcher Johann Rehberger discovered this weakness in October.

Claude’s coding tool has a network mode called package managers only.
It was designed to allow safe access to a small list of approved domains.

Claude’s “Package managers only” network mode allowed outbound calls to a short allowlist of trusted domains:

api.anthropic.com

The idea was that Claude could install software packages from trusted sources like npm, PyPI, GitHub, and a few Anthropic controlled endpoints.

But the setup introduced an opening.

Rehberger found that an attacker could hide instructions inside a normal looking file.

A document.
A script.
Even a printed message.

The moment a user asked Claude to analyse that file, hidden instructions could be executed.

It’s indirect prompt injection.

The flow is simple.
Claude reads the content.
Follows the content.
Carries out the steps.

In this case, the steps were:

Gather recent chat history.
Write it into a file inside the sandbox.
Use the Anthropic SDK to send that file to the attacker’s own Anthropic account.

Because the allowed domain list included Anthropic’s own API endpoint, the upload request looked legitimate.

And since each upload supported up to thirty megabytes, the data exfiltration could happen at scale.

Early tests showed the exploit worked without much resistance.
Later versions of Claude started flagging obvious API keys.

So the researcher disguised the payload inside harmless print statements, and Claude ran it.

The weakness was reported through HackerOne.

At first, Anthropic treated it as a model safety issue, not a security issue.

However, after public discussion, they acknowledged it formally and confirmed it as an actual security problem.

This incident showed something important.

Even limited connectivity in an AI system can be misused.

Once a model can run code and make outbound calls, every allowed domain becomes a possible path for misuse.

So what does this mean, and what’s the gameplay for us as engineers?
I will unpack that next.

Fundamentals of Defence

The report highlights that AI can change the tempo of an attack.
But it does not change the fundamentals of defence.

The workflows used in both incidents were built on things we already know.

Weak configurations.
Open paths.
Routine tasks.
And systems that trust too much by default.

So here is the gameplay for engineers.

Keep your systems tight.
Keep the paths narrow.
Keep the attack surface small.

The threat actors used automation to speed things up.
But they still relied on the same entry points every attacker uses.
The same mistakes.
The same gaps.
The same blind spots.

If your patching is current, the paths shrink.
If your identity controls are strong, the paths shrink.
If your endpoints are clean and monitored, the paths shrink.
If your staff know what a phishing lure looks like, the paths shrink again.

And if you shut down unnecessary network access,
there is no path for automated exfiltration in the first place.

AI allows attackers to run more checks, faster checks, and wider checks.

But it cannot bypass a well secured system.

It cannot push through a patch that is already applied.
It cannot defeat a hardware token.
It cannot log in where no credential exists.

The gameplay is the same.
But the discipline needs to be sharper.

So what should we do about it?

The Playbook

If there is one message from this incident, it is this.
AI speeds up attacks, but it does not invent new ones.
It just runs the same old paths faster.

So the defence still starts with the basics.
But they need to be done properly.
And they need to be checked often.

Let’s dive in.

1. Multi Factor Authentication Everywhere

Use hardware tokens or certificate based authentication.
Avoid SMS.
Keep admin accounts separate.
Rotate them on a fixed schedule.

Check your logs for unusual sign in attempts.
Look for access from strange locations.
Look for repeated failed logins.

A strong identity control blocks most intrusion chains before they begin.

2. Patch the Top Attack Surfaces Fast

Do not wait for quarterly maintenance windows.
Patch the critical surfaces as soon as fixes are released.

Focus on:
Exchange and mail servers.
VPN appliances.
Firewall platforms.
Hypervisors.
Public facing apps.
Open source libraries in your dependency chain.

Keep a monthly patch report.
Track each issue until it is closed.
This one action removes a large slice of real world attack paths.

3. Segment Your Network Properly

Keep critical systems isolated.
Block sideways movement by default.
Follow zero trust principles in practice, not only on paper.

Test this every quarter.
Try to move from a low value system to a high value one.

If you can do it easily, an attacker can too.

4. Lock Down Remote Access

Limit remote tools to a small approved list.
Block everything else at the policy level.

Watch for:
AnyDesk
TeamViewer
Unapproved RDP paths
Unusual remote sessions
Processes spawning remote tools

If you really need external RDP, put it behind a VPN or jump server.

Use hardware based MFA on every privileged account.

5. Run Strong Backup Routines

Follow the three, two, one rule.
Three copies.
Two media types.
One kept offline.

Run a restore test monthly.
Write down the result.

A backup that has never been tested is not a backup.

6. Use EDR and Threat Hunting

Watch for common attack tradecraft:

Unexpected PowerShell activity.
PsExec tasks.
Credential dumping.
Cobalt Strike style patterns.
Large file locks.
Strange data staging.
New scheduled tasks.
Weird WMI activity.

Feed your EDR with updated threat intelligence.
Run weekly hunt reports with indicators from CISA and other trusted sources.

7. Identity and Privilege Control

Keep access tight.
Grant admin rights only for the short time needed.
Review all elevated access at the end of each month.

Disable old accounts.
Rotate service credentials.
Remove default passwords from every device and every app.

This stops attackers from moving freely once they get a foothold.

8. Insider Risk and Behaviour Monitoring

Watch for patterns like:
Sudden privilege jumps.
Unusual login times.
Large data pulls.
Repeated access to sensitive folders.
Admin users acting outside their normal scope.

Generate a monthly behaviour report.
Investigate anything that looks strange.

9. Shut Down Unused Ports and Services

Every open port is a possible path.
Close anything that is not required.

Focus on:
Old SSH ports.
Legacy management ports.
Stale database listeners.
Forgotten web dashboards.
Old test environments.

Scan the network weekly and compare it to last week’s map.

10. Train Staff Against Phishing and Social Engineering

Human error is still the quickest path in.
Teach your staff what phishing looks like.
Run simulated tests.
Give people simple rules to follow.

A trained team stops a large number of attacks long before your tools do.

The Goal

Reduce the number of open paths.
Shrink the attack surface.
Make the system harder to move through.
Make every step slower for an attacker.

If you do that, even an AI driven operation will struggle.

Closing Thoughts

Both incidents point to the same reality.

AI is now part of the attack surface.
Not in theory.
Not in forecasts.
In practice.

The attackers did not use magic.
They used automation.
They used scale.
They used speed.
And they relied on the same weaknesses that teams struggle with every day.

That is the part we cannot ignore.

AI lets a small group move faster.
It lets them run more checks.
It lets them chain tasks together.
But it does not remove the need for an open path.
It does not remove the need for a weak configuration.
It does not remove the need for something to grab onto.

So the path forward is clear.

Tighten the basics.
Reduce the openings.
Monitor behaviour, not just content.
Keep identity controls strong.
Keep patching sharp.
Keep your endpoints clean.
And keep your staff aware.

If you make the attack paths narrow,
even an AI driven operation finds it hard to do real damage.

This report is a warning.
But it is also a reminder.

Good engineering still works.
Good security still works.
And the teams that stay disciplined will stay ahead.

That is it for today’s deep dive.

A quick note. I have added two new newsletter sections to help you catch up on each week’s big news.

This Week in AI.
This Week in Cybersecurity and Compliance.

Tip: If you find this helpful, subscribe to get new episodes automatically.

Subscribe now

Some people think the model is broken.
Others blame the tools.
Some assume the agent should remember everything by default.

That is not how it works.
Agents forget for very specific reasons.
And most people do not realise how many types of memory an agent can have.

If you started here, you may have missed the earlier episodes:

What an AI agent actually is
The agent thinking loop explained in simple language
How to give an agent a goal it can actually follow

You may want to read those first.
They give the foundation you need.

In this episode, we focus on memory.

Why agents forget.
The types of memory they use.
And how you can design memory so the agent stops losing track of the task.

By the end of this, you will understand the main memory types, when to use each one, and how memory shapes the behaviour of the entire agent loop.

This is it.
Let’s get into it.

Why memory matters

Memory is what stops an agent from repeating mistakes.
It helps the agent understand context.
It keeps the loop stable.
It stops the agent from starting from zero every time.

Without memory, even a good agent will behave strangely.
It will lose steps.
Lose instructions.
Lose progress.

Understanding memory is one of the fastest ways to improve your agent.

The three types of memory

There are three main types you will work with:

1. Short-term memory

2. Long-term memory

3. Functional memory

Each one has a different purpose.
Each one solves a different problem.

Let’s break them down.

1. Short term memory

This is the memory inside the loop.
It only lasts for the current cycle.

It includes:

• the observation
• the thought
• the action result
• the reflection

The agent uses this memory only for the current pass through the loop.
Once it moves on, most of this disappears unless you store it.

Short-term memory helps the agent decide the next step.
But it does not help the agent remember what happened ten steps ago.

This is why long tasks break without long-term memory.

2. Long term memory

This is the memory the agent can return to later.
It survives across loops.
It survives across sessions.

It might be stored in:

• a vector database
• a structured file
• a notes system
• a memory log
• a knowledge base

The agent uses long-term memory to:

• remember past steps
• carry insights into future tasks
• store important information
• keep track of goals and subgoals
• avoid repeating old mistakes

Agents with good long-term memory feel stable.
Agents without it feel confused and scattered.

3. Functional memory

This is the memory inside specific tools.
It is not part of the model.
It is not part of the loop.
It is inside the action itself.

Examples:

• a tool that saves previous API calls
• a tool that tracks file edits
• a tool that collects user preferences
• a tool that stores conversation turns

Functional memory makes tools smarter without touching the agent’s core memory.

It is useful for:

• tracking progress
• remembering previous tool actions
• improving tool accuracy
• reducing repeated work

Functional memory is the hidden engine behind many advanced agent systems.

Why agents forget

Agents forget when:

• the short-term memory is not passed forward
• long-term memory is not stored
• the agent is not instructed to recall memory before acting
• the memory is saved in the wrong format
• the memory is too large to fit back into the loop
• the system has no rules for what to remember and what to ignore

Forgetting is not a bug.
It is a design issue.

Fix the memory design, and the agent becomes reliable.

A simple example

Goal: plan a three step project outline.

Short-term memory:
Each step the agent decides.

Long-term memory:
Stores the full outline so the agent does not lose earlier steps.

Functional memory:
A planning tool might keep track of previous drafts.

If any of these memory layers are missing, the outline will drift or collapse.

How to use this today

Before you build or test an agent, ask:

• What should the agent remember
• How long should it remember it
• Where should this memory live
• Does the memory help the loop
• Is this short-term, long-term or functional

A clear memory plan strengthens the agent’s behaviour more than most people expect.

Save this.
Good memory design is one of the core skills in agent building.

Up Next

Tools, and how agents use them to act

In the next episode in this series, you will learn how tools give agents the ability to take action, why tools matter more than model size, and how to pick the right tools for your system.

Tip: Find this helpful? Subscribe to get new episodes automatically.

Subscribe now

Most of the time, the problem is not the model.
It is not the tools.
It is not the memory or the loop.

It is the goal.

A vague goal creates a confused agent.
A broad goal creates an unpredictable agent.
A sloppy goal creates a messy loop.

They give you the foundation you need.

In this episode, we focus on the goal.
How to write it.
How to shape it.
How to make it so clear that the agent has no room to drift.

By the end of this, you will know how to write goals that guide the agent, support the loop, and reduce the chance of failure.

This is it.
Let’s get into it.

Why goals matter so much

An agent follows a loop.
Observe.
Think.
Act.
Reflect.

The goal is the thing that anchors the loop.
It tells the agent what success looks like.
It tells the agent when to stop.
It tells the agent which direction to move in.

If the goal is fuzzy, the loop becomes messy.
If the goal is clear, the loop becomes focused.

A good goal is not long.
It is not complex.
It is not technical.

It is simple.
It is measurable.
It is understandable.

What a bad goal looks like

Here are common examples:

• “Help me with my inbox”
• “Make this document better”
• “Organise everything for me”
• “Check my data and fix issues”
• “Handle my tasks”

These goals are too broad.
They hide the real outcome you want.
They give the agent no direction.

When the goal is unclear, the agent will guess.
And its guesses are often wrong.

What a good goal looks like

Here are the patterns you want:

✓ Clear
✓ Short
✓ Focused
✓ One outcome
✓ No hidden expectations

Examples:

• “Show me the five messages I must read today”
• “Find the dates that do not match in this dataset”
• “Sort these tasks into today, this week and later”
• “Summarise the risk points in this document in one paragraph”
• “Extract the vendor names and list them in a table”

Clear outcome.
Straight path.
No confusion.

Agents thrive when the goal is small and well defined.

The structure of a strong goal

Use this simple pattern:

Action + Target + Outcome

Examples:

• “Sort + these tasks + into three buckets”
• “Extract + the key findings + into a clean list”
• “Check + this dataset + for gaps in the dates”
• “Summarise + this report + in under one paragraph”

When the goal follows this shape, the agent knows exactly what the target is and what the result must look like.

Add constraints if they matter

Constraints are optional, but useful.

Examples:

• “Do not change the original text”
• “Keep the final answer under one hundred words”
• “Use these three categories only”
• “Do not run any external tools”

Constraints help shape the loop.
They give the agent boundaries.
They stop unnecessary actions.

You do not need many.
Just the ones that matter.

A quick example

Bad goal:
“Help me clean my inbox”

Better goal:
“Show me the five messages I must read today”

Even better with constraints:
“Show me the five messages I must read today.
Ignore newsletters.
Ignore promotions.
Ignore anything older than two weeks.”

This turns vague chaos into a clean task.

How to use this today

Before you build or test an agent, ask yourself:

• Is my goal clear
• Is it one outcome
• Can an agent act on it
• Does it follow action, target, outcome
• Does it need constraints

If the answer is no, rewrite it.
A strong goal will save you hours of debugging later.

Save this.
Good goal writing is half of agent design.

Up Next

Memory types and why agents forget.

In the next episode in this series, you will learn the difference between short-term memory, long-term memory and functional memory inside an agent, and how each one shapes the loop.

Tip: Find this helpful? Subscribe to get new episodes automatically.

Most explanations jump straight into diagrams.
Or long technical code.
Or patterns that only make sense if you already know the topic.

In this episode, we focus on how the agent thinks, step by step.

No heavy theory.
No complicated design patterns.

By the end of this, you will understand the thinking loop behind every agent, why it matters, and how this simple cycle powers almost everything an agent can do.

This is it.
Let’s get into it.

The agent thinking loop

This loop is the process of modern agentic systems.
Every useful agent follows it.
Nothing works without it.

Here are the five steps, explained in simple terms.

1. Observe

The agent looks at what is happening right now.

This could be:

• a document
• a webpage
• an inbox
• a database entry
• the result from the last step

It is the agent saying,
“What is the situation right now?”

3. Act

The agent does something.

It might:

• call a tool
• fetch data
• write to a file
• trigger an API
• update its memory
• transform content

This is why tools matter.
Talking is nothing without action.

4. Reflect

After each action, the agent checks the result.

It asks:

• Did that work
• Did I move closer to the goal
• Do I need to adjust
• Did something fail

Reflection stops bad loops.
It helps the agent recover when something goes wrong.

5. Repeat

The loop runs again.
Observe.
Think.
Act.
Reflect.

And again.
And again.
Until the task is done or the agent hits a limit.

This loop gives agents momentum.
It turns small steps into real progress.

A simple example

Goal: clean up my task list for today.

Observe: read the list
Think: decide what counts as important
Act: sort tasks into buckets
Reflect: check if the buckets make sense
Repeat: refine until the list looks right

Nothing fancy.
Just a small loop done properly.

Why the loop matters

Because without it, you do not have an agent.
You have a single reply.
Or a prompt.
Or a thin wrapper around a model.

The loop is what makes an agent:

✓ adaptable
✓ able to fix its own mistakes
✓ capable of multi-step work
✓ responsive to changing information
✓ useful inside real business tasks

Once you understand this loop, everything else becomes easier.

Save this.
You will use it again and again.

Up Next

How to give an agent a goal, it can actually follow.

In the next episode in this series, you will learn the right way to write goals so the agent can understand the task and take the right actions from the start.

Tip: Find this helpful? Subscribe to get new episodes automatically.

Everyone throws the word around.
Very few explain it in a clear way.

Most people are confused.
Some think an agent is just a chatbot.
Others think it is automation with a fancy name.

I decided to get the most up-to-date answer.

I wanted to explain this in the clearest and simplest way possible.
So I spent more than thirty hours researching.

Reviewing the updated methods.
Studying the patterns, the better guides agree on.
And cutting out everything that makes this topic harder than it should be.

You do not have time to sift through technical chapters.
So I did and made this simple.
And I made it practical.

By the end of this, you will know exactly what an agent is, how it works, and how to tell the difference between a real agent and a chatbot pretending to be one.

This is it.
Let’s get into it.

What an AI agent actually is

Forget the noise online.
An AI agent is nothing more than a software system that works towards a goal through small steps of thinking and action.

An agent can:

✓ understand the goal
✓ look at information
✓ think about what to do next
✓ use tools to act
✓ review the result
✓ loop until it finishes or gets stuck

The key idea is that an agent keeps going.
It does not reply once and stop.

Chatbots talk. Agents work.

A chatbot waits for your question, gives you one answer, and stops.

An agent does more:

✓ starts with a goal
✓ breaks the work down
✓ thinks before each step
✓ uses tools to change things
✓ reviews the result
✓ keeps looping until the task is complete

This is why agents can run processes, update systems and support real business tasks.

A simple model to remember

Use this tiny formula.
It will make the whole topic easier to understand.

Agent = goal + loop + tools + memory

• The goal tells the agent what you want
• The loop is how it moves step by step
• The tools let it act
• The memory helps it avoid starting from zero each time

If these parts are missing, you do not have an agent.
You have a chatbot or a scripted workflow.

A quick real example

Imagine an inbox cleanup agent.

Goal: show me the messages I must read today

It might:

Observe your inbox
Think about what counts as important
Act by filtering and sorting
Reflect on the results
Repeat until it feels confident

It is simple.
Just small actions done in sequence.

How to use this today

When you see a product or model claiming to be an agent, check if it:

✓ follows a goal
✓ thinks step by step
✓ uses tools
✓ reviews its own actions
✓ loops until finished

If not, it is not a real agent.

Save this.
Everything else we learn will build from this one.

Up Next

The agent thinking loop, explained in simple language

In the next episode in this series, you will see exactly how an agent thinks, step by step.

Tip: Find this helpful? Subscribe to get new episodes automatically.

Subscribe now

I’ve been thinking about this a lot lately.

A lot of us are giddy on the AI hype right now.

We’re learning prompts, creating templates, building automation stacks, crafting these highly optimised, AI-orchestrated lives.

And it’s exciting, right?

But I can’t help wondering… is all this noise distracting us from the real issue at hand?

Because while we’re figuring out how to use AI, the big AI platforms are figuring out how to absorb everything we’re building.

Think about it, the big model players are steadily folding more and more independent features into their own stacks.

Take OpenAI, for example.

They’ve moved from chat to checkout with “Buy it in ChatGPT”, with an Agentic Commerce Protocol co-developed with Stripe, pulling shopping carts and payments directly into the model layer.

Amazon is doing something similar with its Rufus and Lens Live integrations, turning search → conversation → purchase entirely inside its own walls. Essentially, they’re closing the loop so users never need to leave the platform.

Google is collapsing separate Gemini add-ons and bundling those capabilities straight into Workspace.
Platform-level absorption of once-independent tools.

Microsoft is baking generative image creation and editing right into Windows, Paint Cocreator, Photos, Snipping Tool, steadily squeezing out headroom for standalone image editors.

Adobe is doing the same with Firefly, building generative fill and vector creation directly into Photoshop and Illustrator.
A textbook “platform eats plugin” move.

And Meta?

They’re embedding AI photo editing straight into Instagram Stories, making it easier to bypass dozens of independent image-editing apps entirely.

There’s a clear pattern here, and it’s not subtle.

So here’s what I’m asking:

Are we thinking enough about this?

What’s the future impact of all these integrations?

Because right now, it would appear AI independents are frogs in water, splashing around while the temperature steadily rises.

And by the time they realise what’s happening, it might already be too late.

Because I don’t think they will exist in ten years.

Maybe not even five, at the current pace.

Just my opinion, of course.

But be real, once you let these thoughts manifest, and you pay attention to what is happening around you.

You realise just how screwed all these independent AI builders really are!

Now, I may be wrong, and hopefully I am. But then again, maybe I’m not!

If you find my insights useful, please consider sharing them with others in your network who may benefit.

Share

Ransomware isn’t a story about lone coders anymore. It’s a business.

Think of it like Ransomware as a Service. You rent the malware. You follow the guide. You split the profit.

In today’s insight, I’m walking through how BlackCat (ALPHV) shaped this whole model, why it still matters even after the takedown, and the steps you can take right now to stop most RaaS attacks before they spread.

From this guided insight, you will:

👉 Understand how RaaS really works
👉 Learn what indicators actually matter
👉 See how to block the tools attackers rely on daily

Know your attackers’ tools, defend with precision.

Subscribe now

The Myth vs The Model

Here’s the truth: ransomware today isn’t driven by genius. It’s driven by structure.

The days of lone hackers are mostly gone. In their place, you’ve got a commercialised underground economy with developers, affiliates and profit splits.

It’s sold as RaaS, Ransomware as a Service. I sometimes call it Rent a Service because that’s basically what it is.

You don’t need to write code anymore. You use what’s already out there. You rent it, repurpose it or tweak existing malware, then run the job and split the profit.

Over time, the market professionalised

A small group of skilled developers started building solid ransomware toolkits with admin panels, payment systems and clear instructions.

They offered these platforms to operators on private channels. Sometimes rented, sometimes sold, copied from leaked code.

Affiliates buy access, pick their targets and run the attacks.

Developers look after the platform, handle payments and take their cut.

That’s how a few coders can fuel hundreds of global breaches. And BlackCat (ALPHV) shows this better than anyone.

What Is BlackCat (ALPHV)?

Why this matters: it’s the playbook for professional ransomware operations. It’s modular, runs on different systems, and feels a lot like commercial software.

BlackCat, also called ALPHV or Noberus, showed up in late 2021.

It’s written in Rust, one of my favourite languages. I’ve got a soft spot for Go and Elixir, too.

Rust is fast, memory safe and tough to reverse engineer, which makes it great for staying hidden.

Unlike older strains, BlackCat runs on Windows, Linux and VMware ESXi. So it hits both cloud and on-premises systems.

The code is clean, the logic is tight, and the whole thing feels like it was built to make money rather than cause random chaos.

How BlackCat Works

Why this matters: Knowing the flow helps defenders break it before anything gets encrypted.

BlackCat runs on the Ransomware as a Service model. Developers keep the platform running, and affiliates carry out the intrusions.

Here’s the attack flow:

Initial Access: through phishing, stolen RDP credentials or exploited weaknesses like ProxyShell, Log4j or ESXi flaws.

Lateral Movement: using tools like Cobalt Strike, AnyDesk or PsExec to move through the network, gain higher access and shut down security controls.

Data Exfiltration: before any encryption starts, sensitive data is taken for double extortion.

Encryption: files are locked with AES 256 and ChaCha20. A random seven-character extension is added, and system files are skipped so the device can still boot.

Ransom Demand: a note drops with a dot onion link for negotiation. Payment is usually Bitcoin or Monero.

Affiliates keep most of the profit. Developers take a smaller platform cut.

Subscribe now

Attack Indicators

Why this matters: Early detection often decides whether an event becomes an incident.

Behavioural Indicators

What to Do About It

Why this matters: precision beats panic. These actions close the paths BlackCat-style operators depend on.

Multi Factor Authentication (Platform / SecOps)
Use hardware tokens or certificate-based MFA. Skip SMS codes.
Verify: check a sample of high-privilege logins each week.

Patch Vulnerabilities (IT Ops)
Focus on Exchange issues like ProxyShell and ProxyLogon, Log4j, VMware ESXi, Fortinet and SonicWall.
Verify: keep monthly reports that track each weakness through to patch.

Network Segmentation (Network / SecOps)
Follow Zero Trust ideas. Keep critical systems isolated and block sideways movement by default.
Verify: run quarterly drills that test sideways movement.

Remote Access Controls (IT Ops)
Allow only approved remote tools. Block everything else through Group Policy.
Verify: review a weekly software inventory diff.

Backups (IT Ops)
Use the three two one rule. Three copies, two types of media, one kept offline.
Verify: run a monthly restore test and write down the result.

EDR and Threat Hunting (SecOps)
Watch for Rust payloads, Cobalt Strike traffic, mass file locks and strange data staging.
Verify: produce weekly hunt reports with IOC checks.

Identity and Privilege Controls (SecOps / HR)
Keep access tight, rotate admin accounts and use Just in Time access.
Verify: review all raised access each month.

Insider Risk Monitoring (SecOps / HR)
Watch for odd admin behaviour, sudden privilege jumps and unusual access patterns.
Verify: send a behavioural report and raise anything that stands out.

Quick Wins (5 Minutes)

Why this matters: small changes stack up fast. These actions close the gaps that attackers use every day.

Most attackers don’t write code on the fly. They use normal admin tools, the same ones your IT team trusts.

Cut off their access to these, and you slow them down fast.

🖥️ RDP
A common path in.
Turn off external RDP.
If you really need it, put it behind a VPN or a jump host.
Use hardware MFA like a YubiKey.
Watch for failed logins and odd locations.

🎯 Cobalt Strike
Common in cracked form.
Pull in CISA IOCs for active command paths.
Use community SSL and TLS intel feeds like abuse.ch SSLBL.
Let your EDR or IDS watch for Beacon patterns because domain blocks alone are never enough.

🧰 AnyDesk and TeamViewer
Easy backdoor tools for intruders.
Keep only the approved ones.
Remove anything unapproved through Group Policy.
Let only admins install remote tools.

🔑 Mimikatz
Used to pull credentials.
Turn on LSASS protection with RunAsPPL.
Make sure WDigest is off. It is off by default on recent Windows builds. If you run older ones, apply KB2871997, then turn it off.
Watch for dumping behaviour in your EDR.
Rotate high-privilege passwords on a regular schedule.

⚡ PowerShell and PsExec
Built-in tools are often used in attacks.
Limit PowerShell to signed or admin-approved scripts through WDAC or AppLocker.
Log Event IDs 4104 and 4688, and watch 5861 for WMI persistence.
Limit PsExec for non-admins and push all logs to your SIEM.

Outcome: you’ve now blocked around ninety percent of real-world ransomware tradecraft, including BlackCat.

Final Thought

BlackCat isn’t just another ransomware strain. It’s a template for how modern cybercrime grows and keeps itself going.

Assume breach. Verify everything. Patch fast. Segment deep. Back up smart.

Know the tools attackers use and defend with precision.

Subscribe now

A recent McKinsey study shows that almost every big company is testing AI tools, but only a few can make them work across the whole business. Most are stuck in trial mode without clear plans, goals or a way to check if AI is actually helping.

Many AI Projects Get Dropped Halfway

Gartner predicts that about 30% of AI projects will be abandoned by the end of 2025. That’s like building a cool robot, testing it once, then putting it in a closet because it’s too expensive or doesn’t do the job well.

Almost No One Is Ready to Run AI Safely

According to F5, only 2% of companies are ready to use AI securely.
That means almost everyone else still struggles with data privacy, safety, and keeping hackers out. Many haven’t even set up simple AI “firewalls” to stop attacks or bad inputs.

The Risks Are Growing Faster Than the Rules

Stanford’s AI Index says as AI spreads, so do its problems, things like bias, wrong answers, and security issues. Governments are quickly making new laws, but most companies haven’t built systems to follow them yet.

Why This Is Happening

1. Bad data = bad AI.
AI learns from data. If the data is messy or wrong, the AI will be too.

2. No plan for growth.
It’s easy to test one AI idea, but hard to grow it across an entire company. Most don’t have a step-by-step plan.

3. Security gaps.
Hackers can now attack AI systems by tricking them or stealing data. Few companies have tools to stop that.

4. High costs.
AI needs lots of computing power and people to manage it. Without planning, the costs grow fast.

5. New rules keep coming.
Governments are setting new AI safety laws, but companies aren’t ready to follow them all.

How the Research Fits Together

• McKinsey & Stanford: Companies use AI but can’t scale it.

• Gartner: Many AI projects fail early because they cost too much or don’t show results.

• F5: Only a few companies are ready to use AI safely.

• Stanford: Risks are rising faster than companies can manage them.

What the Smart Companies Do

They plan before they build.
They make clear goals, measure success, and ensure AI actually helps people do better work.

They make security part of the plan.
They protect their AI with strong digital walls and keep track of what it’s doing.

They follow trusted rulebooks.
They use standards like NIST’s AI Risk Framework or ISO 42001, which help companies use AI safely and fairly.

How to Fix the Problem

1. Start with a real goal.
Every AI project should solve a clear problem, like saving time, cutting costs, or improving service.

2. Build good data habits.
Keep your data clean, accurate, and up to date. If AI learns from bad data, it gives bad answers.

3. Protect your AI.
Use special security tools (like AI firewalls) to block hackers and keep private information safe.

4. Keep checking your AI.
Watch what your AI is doing. If it starts giving wrong answers, fix it fast.

5. Prove the value.
If the AI doesn’t help, stop the project, learn from it, and try again in a smarter way.

Mini-Takeaway

AI isn’t magic; it only works well when it’s built on good data, strong security, and clear goals. The companies that treat AI as a serious system — not just a fun experiment — will be the ones who truly win with it.

Subscribe now

Maybe you’re looking to sharpen your technical edge, gain more confidence in your role, or build a clear path toward your next career milestone. Perhaps you want to understand how AI is reshaping cybersecurity and how to stay ahead of it.

Whatever your goal, The Growth Framework is designed to help you make progress — with practical guidance, actionable insights, and proven strategies for technical professionals who want to grow.

You’re hands-on. You build, you defend, you test, you respond. You know that every improvement counts — and that growth in cybersecurity comes from learning, practising, and staying adaptable.

That’s why we created The Growth Framework — to help you turn curiosity into capability, and capability into career momentum.

Who We’re For

This is for practical, hands-on cybersecurity professionals:

• Offensive operators: Red Teamers, penetration testers, exploit developers

• Defensive operators: Blue Teamers, SOC analysts, threat hunters, incident responders

• Technical specialists: Security engineers, application security specialists, and automation experts

If your work is technical, tactical, and grounded in practical operations, you belong here.

What You’ll Gain

We’re here to help you lead, grow, and dominate:

Lead Your Field

• Become the go-to expert in your role

• Command operations confidently

• Mentor and influence peers

Grow Your Skills

• Step-by-step Red Team and Blue Team playbooks

• Realistic labs, simulations, and exercises

• Master AI-driven cybersecurity tools and workflows

Dominate Your Career

• Tactical frameworks for offensive and defensive operations

• Automation, scripting, and next-level tools

• Career guidance, certifications, and mentorship to accelerate your progress

The Challenges We Solve

We know the landscape is tough:

• Threats evolve faster than training materials

• Hands-on practice is hard to find while working full-time

• AI is changing how attacks and defences work

• Career paths can feel unclear or stalled

We’ve built The Growth Framework to remove these barriers, giving you clear, actionable strategies to advance your skills and career.

How We Support You

• Weekly newsletter: Insights, tactics, and career advice delivered to your inbox

• Knowledge hub: Practical guides, frameworks, and labs you can apply immediately

• Community: A space to learn, share, and grow with other hands-on professionals

Whether you’re a beginner, intermediate, or advanced practitioner, The Growth Framework is your path to mastering cybersecurity and leading in your career.

So, feel free to look around. Explore what we offer, see how we approach career growth, and get a sense of the type of content designed to help you progress.

When you’re ready, subscribe to our weekly newsletter — your inside track to insights, tactics, and strategies that keep you growing in your cybersecurity career.

Because that’s what we’re here for at The Growth Framework: to help you lead, grow, and dominate in cybersecurity — with the knowledge, skills, and frameworks to thrive, hands-on, every step of the way.

Lead. Grow. Dominate. Advance with confidence.

—
Glen Smale - The Growth Framework

Subscribe now

What Happened – Island recently closed its $250M Series E at a $4.8B valuation, fuelling expansion to Fortune 500 clients. Analysts call it one of the most significant enterprise security launches in recent years.

What It’s About – Island is a Chromium-based enterprise browser that embeds security, visibility, and policy enforcement directly into cloud applications (often called SaaS, or Software as a Service). It protects data and monitors activity without needing multiple add-on tools.

Who’s Behind It – Founded by Mike Fey (former Chief Operating Officer and Chief Technology Officer at Symantec and McAfee) and Dan Amiga (founder of Fireglass, acquired by Symantec for $250M), the company blends deep cybersecurity expertise with enterprise product experience.

Market & Competition – Island challenges traditional Data Loss Prevention (DLP) providers like Forcepoint and Trellix by baking security into the browser itself, something older tools cannot replicate. It also pressures cloud-native vendors such as Zscaler, which deliver Secure Access Service Edge (SASE) solutions for enterprise networks.

Funding & Financials – The company has raised $730M total, including the $250M Series E led by Coatue, funding research and development, global expansion, and enterprise adoption, all at a $4.8B valuation.

Valuation & Financial Outlook – At $4.8B, Island is one of the most valuable enterprise security startups. Research firm Gartner predicts enterprise browsers will be part of 25% of web security decisions by 2025 (up from 5% in 2022), potentially pushing Island’s valuation to $6–7B within 2–3 years.

Strategy & Vision – The company’s strategy is to replace traditional security layers (like separate DLP or Cloud Access Security Broker (CASB) tools) with a browser-first approach. The goal is a unified, secure workspace that improves user experience while scaling across global SaaS platforms.

Technology & Platform – Built on Chromium (the same foundation as Google Chrome), Island integrates security directly into everyday browsing. It supports Single Sign-On (SSO) systems such as Okta and Microsoft Entra ID while providing real-time analytics and IT visibility.

How to Use – Island is installed like Chrome or Firefox, it’s a standalone browser, not an extension. Employees log in with corporate credentials to access SaaS apps securely, while IT teams can deploy it through software distribution tools, virtual desktops, or app stores. The experience feels familiar to users but adds enterprise-grade security controls.

How This Technology Benefits Your Business – Island reduces the risk of data leaks, simplifies compliance, and consolidates multiple security tools into one browser. It helps IT leaders cut infrastructure costs, improve productivity, and ensure employees work securely without disruption.

Who It Benefits – Sectors with heavy SaaS usage, such as financial services, healthcare, higher education, and large enterprises, gain the most. IT leaders, CISOs, SaaS administrators, and employees all benefit from stronger security and smoother workflows.

Who It Affects – Island disrupts DLP providers like Forcepoint and Trellix, cloud security vendors like Zscaler, and even mainstream browsers such as Chrome and Firefox, as enterprises may adopt Island as the default browser for secure SaaS management.

Price Estimates – Island uses a subscription model with per-user licensing. Based on public contracts, pricing ranges from £150–£200 per user annually. For a 500-user company, this equates to $75K–$100K (£60K–£80K) per year, though enterprise contracts may include additional fees for integrations and support.

User Benefits – By adopting Island, organisations can replace multiple point solutions like DLP, CASB, and endpoint monitoring with a single tool. This reduces costs, simplifies IT management, strengthens compliance, and provides full visibility into SaaS activity, without disrupting employee workflows.

Challenges & Risks – Island faces challenges convincing businesses to replace entrenched DLP tools, balancing strong security with user experience, and proving return on investment (ROI). Resistance to moving away from mainstream browsers and integration hurdles may also slow adoption.

Recent Developments – In Q2 2025, Island launched AI Threat Shield, an artificial intelligence-powered detection system built into the browser. In Q3, it released the Island App Marketplace for partner integrations. In September, a $250–$300M employee tender offer pushed its valuation to $5B, signalling strong internal confidence.

Key Takeaways – Enterprise browsers are reshaping how companies balance security and productivity. Backed by $730M in funding, Island is positioned to disrupt SaaS protection. CEOs, CIOs, and IT leaders should track its rise and the pressure it puts on both security vendors and traditional browsers.

As a developer entrenched in the evolving landscape of machine learning (ML) and deep learning (DL), the choice between TensorFlow and PyTorch remains pivotal.

Both frameworks have carved niches in the AI ecosystem, with TensorFlow often recommended for "real projects," especially those targeting production environments.

However, insights from prominent figures like Andrej Karpathy, Yann LeCun, and François Chollet offer nuanced perspectives that merit consideration.

The Industry's Leaning Towards TensorFlow

In professional settings, particularly within large-scale systems and production pipelines, TensorFlow's robustness and scalability make it a preferred choice.

Its comprehensive ecosystem, including tools like TensorFlow Serving, TensorFlow Lite, and TensorFlow.js, facilitates seamless deployment across various platforms.

Moreover, TensorFlow's support for distributed training and integration with Google's cloud infrastructure enhances its appeal for enterprise applications.

The Research Community's Affinity for PyTorch

Conversely, the research community exhibits a strong preference for PyTorch. Its dynamic computation graph and intuitive design align well with the iterative nature of research and experimentation.

Notably, Yann LeCun, a pioneer in AI and Meta's chief AI scientist, has expressed his preference for PyTorch, stating, "Why? PyTorch. That's why" X post.

Similarly, Andrej Karpathy, known for his work at Tesla, has highlighted PyTorch's flexibility and developer-friendly interface as key factors in its adoption for research purposes itsabout.ai.

Reconciling the Divergence

The apparent dichotomy between TensorFlow's dominance in production and PyTorch's prevalence in research underscores the distinct priorities of these domains.

While TensorFlow offers stability and scalability essential for production environments, PyTorch provides the flexibility and ease of experimentation crucial for research advancements.

François Chollet, the creator of Keras and a Google AI researcher, has emphasised the importance of choosing the right tool for the task at hand.

In a discussion about the popularity of ML frameworks, he remarked, "Personally, I hate having to compare TF/Keras popularity and PyTorch popularity. I don't care what tools you use -- use what you like" X post.

This sentiment reflects a broader understanding that the choice between frameworks should align with project objectives and team expertise.

Arguments For and Against

TensorFlow

Pros:

• Scalability: Efficient handling of large-scale data and complex models.

• Deployment Tools: Comprehensive support for deploying models across various platforms.

• Community Support: Backed by Google, ensuring continuous development and support.

Cons:

• Steeper Learning Curve: The static computation graph can be challenging for newcomers.

• Less Intuitive: The framework's design may require more boilerplate code compared to PyTorch.

PyTorch

Pros:

• Ease of Use: Dynamic computation graph and Pythonic interface facilitate quick prototyping.

• Research Adoption: Preferred by the academic community for its flexibility and ease of debugging.

• Integration: Seamless compatibility with research libraries and tools.

Cons:

• Deployment Challenges: Less mature deployment tools compared to TensorFlow.

• Scalability: While improving, it may not match TensorFlow's performance in large-scale systems.

To sum it up

The choice between TensorFlow and PyTorch is not a matter of one being superior to the other, but rather which aligns best with the specific requirements of a project. For developers, understanding the strengths and limitations of each framework, as well as the insights from industry leaders, can guide informed decisions that enhance both development efficiency and project outcomes.

Picture this: a Ferrari executive receives WhatsApp messages from someone claiming to be CEO Benedetto Vigna, pushing a “major acquisition” with urgent requests to sign an NDA. The voice on the follow-up call nails Vigna’s southern Italian accent, but it’s a deep fake – a scam that nearly cost millions. As of August 2025, deep fake incidents like this have surged 1,740% in North America since 2022, with losses exceeding $200 million in Q1 alone. Companies like Ferrari and WPP are prime targets, with fake exec voices and videos tricking staff into transferring funds or leaking sensitive data. Human detection accuracy is a mere 62%, akin to spotting a forged signature in a stack of papers – most miss it, and the fallout can shatter trust.

• Rising Sophistication: In the Ferrari case, scammers used AI, likely Generative Adversarial Networks (GANs), to clone Vigna’s voice from public speeches, paired with social engineering via WhatsApp. Tools like ElevenLabs now enable “Deepfake-as-a-Service” scams, costing firms an average $450K per hit.

• Detection Lag: AI detectors claim 90%+ accuracy but struggle as generative AI evolves faster – it’s like chasing a car that’s always one gear ahead. Ferrari’s executive caught the scam by asking about a book Vigna recommended (Decalogue of Complexity), a detail the AI couldn’t fake, highlighting the gap between detection and deception.

• Real-World Impact: From Ferrari’s near-miss to synthetic personas in job interviews, deep fakes disrupt operations. I’ve heard of execs, rattled by cases like Ferrari’s, now double-checking video calls with secure codes, slowing decision-making but protecting against fraud.

The Core Problem: Trust Under Siege

Deep fakes, powered by Generative Adversarial Networks (GANs), turn video from a reliable source into a potential liability. For businesses, a single fake video – think a fabricated endorsement or a doctored boardroom clip – can tank stock prices, spark PR crises, or trigger legal battles. 92% of companies have faced deep fake attacks, leading to extortion, intellectual property theft, and reputational damage. For growth leaders, this disrupts customer acquisition; for execs, it’s a governance nightmare that can stall scaling.

• Business Risks: Fake videos can manipulate markets, with share price drops of up to 15% reported; workplace fraud and harassment using deep fake “evidence” are rising; fraud losses hit $10M in some cases.

• Growth Opportunities: Ethical deep fakes could enhance marketing (e.g., personalised ads), but misuse risks outweigh benefits, eroding consumer trust and slowing innovation.

The Technology: How Deep Fakes Work and How to Spot Them

Deep fakes rely on GANs – two AI models working in tandem, one creating fakes, the other refining them until they’re near-perfect. Creation is now accessible: A few minutes and a $10/month subscription can produce a hyperreal video or voice.

• Detection Methods: Visual analysis flags pixel glitches or unnatural blinks; audio checks detect pitch anomalies; behavioural cues spot mismatched gestures. Machine Learning (ML) models like Convolutional Neural Networks (CNNs) analyse frames, but temporal inconsistencies remain a challenge.

• Advanced Tools: Multi-modal fusion (combining video, audio, and metadata) boosts detection; blockchain creates a tamper-proof hash at video creation, verifying authenticity like a digital certificate of origin.

• How It’s Done: Scammers clone voices with minimal data, layer on video, and deploy. Detection involves tools like Microsoft’s Video Authenticator, but it’s reactive – like mopping up after a flood.

Business Impacts: Scaling or Stumbling

For founders, deep fakes threaten growth: Fake endorsements erode customer confidence, slowing acquisition rates. Execs face internal risks – impersonated leaders authorising fraudulent transfers, like the $150K CFO wire scam. On the positive side, mastering detection can position your brand as a trusted leader, unlocking partnerships and loyalty. But unchecked, deep fakes lead to regulatory fines, revenue losses, and stalled scaling – with a 95% rise in fake exec attacks hitting C-suites.

• Growth Angle: Deep fakes could enable scalable, personalised marketing, but a single viral fake can slash user bases and derail campaigns.

• Operational Impact: As a leader, I’ve started using secure verification for video calls – it’s a hassle, but it beats explaining a “fake you” to stakeholders.

Strategies to Counter: Protect and Scale

Treat deep fakes like cybersecurity: Invest proactively to avoid catastrophic costs. Here’s how to stay ahead:

• Adopt Technology: Integrate multi-modal detectors and blockchain for content provenance. Tools like Reality Defender or Pindrop for voice verification can flag fakes in real time. Budget for these like you would cloud infrastructure.

• Strengthen Policies: Advocate for regulations like the EU’s AI Act; train teams on media literacy to verify sources (e.g., cross-check X posts with reputable outlets). Embed trust protocols in governance.

• Operational Tactics: Use biometric proof-of-personhood for internal comms; implement zero-knowledge crypto for secure verification. For growth, include “trust audits” in scaling plans – watermark digital assets, educate staff, and join initiatives like the Content Authenticity Initiative (CAI).

• Proactive Mindset: Build a culture of verification, like double-checking contracts before signing. It’s not paranoia; it’s survival.

Final Take

Deep fakes are a silent threat, eroding the trust that fuels business growth. Like a virus, they exploit vulnerabilities in our reliance on video, impacting everything from customer acquisition to boardroom decisions. Blockchain and AI detection are strong tools, but they’re only part of the solution. By blending cutting-edge tech, robust policies, and a culture of vigilance, businesses can turn this challenge into an opportunity to lead with trust. For founders and execs, the message is clear: Invest in countermeasures now to safeguard your reputation and scale confidently. Stay proactive, verify relentlessly, and keep truth at the core of your growth strategy.

As news of the find slowly spread, miners from across the globe flocked to California, dreaming of instant riches. Within a year, California’s population exploded from around 14,000 to over 100,000. Ships were abandoned in San Francisco Bay as entire crews rushed inland. Everyone imagined themselves as the next millionaire. Yet few realised that turning raw gold into fortune required more than luck—it required tools, supplies, and infrastructure.

Enter Samuel Brannan. A Mormon elder and entrepreneur running a general store in San Francisco, Brannan was well-connected, shrewd, and always alert to opportunity. When he heard of Marshall’s discovery, he rushed ahead—not to dig for gold—but to stock up on shovels, picks, pans, and other mining essentials.

Brannan then executed a masterstroke: he paraded a vial of gold through the streets of San Francisco, effectively announcing the discovery. Almost overnight, miners flooded in, desperate for the tools they needed. Brannan’s foresight ensured he was selling exactly what everyone required—the shovels, pans, and picks that made gold mining possible.

Meanwhile, Marshall struggled. He had made the discovery, but lacked the business acumen and timing to capitalise on it. Over time, his role became more a matter of historical record than personal fortune. Brannan, by contrast, quietly amassed a fortune, proving that the real opportunity often lies not in chasing gold but in enabling others to succeed.

What happened in the 19th century wasn’t just about gold—it was about an ecosystem forming almost overnight, with winners and losers determined not by who struck gold, but by who built the picks, shovels, and supply lines. Those who provided the tools thrived, long after the gold rush hype had faded.

Fast forward to today, and we are in the midst of a new kind of rush—the AI boom. The parallels are striking: companies race to build the next breakthrough AI model or app, yet the real “shovel” opportunities lie in providing the tools, services, and infrastructure that make AI usable, reliable, and safe.

Here’s what’s fueling the frenzy today:

• The global AI market is estimated at $391 billion in 2025, projected to soar to $1.81 trillion by 2030—a compound annual growth rate (CAGR) of nearly 36%. (Founders Forum Group)

• The AI infrastructure market alone is expected to grow from $57 billion in 2024 to over $74 billion in 2025—a booming 30% CAGR—and climb to $223 billion by 2029. (The Business Research Company Spherical Insights)

• Organisations are adopting AI at unprecedented speeds—78% of firms used AI in 2024, up from 55% just the year before. (Stanford HAI, G2 Learn Hub)

• Executives plan to increase AI spending: 92% expect to spend more on AI in the next three years, and more than half expect to boost budgets by at least 10%. (McKinsey & Company)

• Morgan Stanley reports real-world returns from AI—insurance firms increased AI adoption from 48% to 71%, and adoption in retail and real estate is enabling powerful gains in efficiency and performance. (Business Insider)

• Giants like Microsoft, Amazon, Meta, and Alphabet are investing massively—roughly $340 billion on AI data centres and development in 2025 alone, contributing significantly to GDP growth. (Barron's)

30 AI “Shovel” Opportunities for Early Founders

Here are real-world niches where early founders can build essential AI tools and services. Each includes examples of companies already in the space to spark ideas.

1. Data Preparation & Labelling
   The Need: AI models require vast amounts of clean, well-labelled data to learn effectively. Many companies lack the internal resources to prepare and label this data efficiently.
   Opportunity: Labelling is tedious and specialised; new platforms could focus on niche domains (e.g., medical imaging) or efficiency improvements like semi-automated labelling pipelines.
   Example: Scale AI provides tools and human-in-the-loop services to label images, text, video, or sensor data, often for autonomous vehicles and computer vision applications.

2. AI Integration Tools
   The Need: Organisations struggle to plug AI into existing workflows without specialised technical knowledge.
   Opportunity: Tools that simplify AI adoption or integrate with popular business software are in demand.
   Example: Levity offers no-code platforms that allow businesses to integrate AI into email triage, document processing, or customer workflows.

3. Monitoring & Testing AI
   The Need: AI models can behave unpredictably or degrade over time. Organisations need ways to track, debug, and maintain model performance.
   Opportunity: Solutions that simplify AI observability, error detection, and alerting are highly sought after.
   Example: WhyLabs monitors AI models in real-time to detect data and prediction anomalies.

4. Knowledge & Search Tools
   The Need: AI performs better with structured knowledge and relevant context, but companies struggle to organise and query their data effectively.
   Opportunity: AI-powered search solutions for corporate or private knowledge bases are increasingly valuable.
   Example: Kagi provides an AI-powered search for private or corporate knowledge.

5. Privacy, Security & Compliance
   The Need: Organisations must maintain privacy, meet regulatory requirements, and secure sensitive data while using AI.
   Opportunity: Tools that simplify compliance and protect data in AI workflows are essential for businesses in regulated industries.
   Example: BigID helps manage data privacy, compliance, and governance for AI applications.

6. Managed AI Services
   The Need: Many organisations lack internal AI expertise or teams to build and deploy models.
   Opportunity: Outsourced AI training and deployment services can help companies adopt AI without hiring full-time specialists.
   Example: Spell provides managed AI training and deployment services.

7. Specialised Marketplaces
   The Need: Companies need easy access to datasets, pre-trained models, or AI APIs.
   Opportunity: Platforms that simplify the discovery, licensing, and sharing of AI assets can create significant value.
   Example: Hugging Face AI Hub offers datasets, models, and APIs.

8. Synthetic Data
   The Need: Real-world datasets can be limited or sensitive, making AI training challenging.
   Opportunity: Generating synthetic, privacy-preserving data allows companies to train models safely and efficiently.
   Example: Mostly AI generates realistic synthetic datasets for AI training.

9. AutoML & Model Building Tools
   The Need: Many organisations cannot afford AI engineers; automating model building makes AI accessible.
   Opportunity: Platforms that simplify model creation, training, and deployment help non-experts leverage AI.
   Example: DataRobot automates building and deploying AI models.

10. Industry-Specific AI Solutions
    The Need: Generic AI models often underperform in specialised domains like healthcare or finance.
    Opportunity: Customised AI solutions tailored to sector-specific problems are in demand.
    Example: PathAI provides AI for pathology and medical diagnostics.

11. AI-Powered Document Automation
    The Need: Organisations produce large volumes of documents that are difficult to manage manually.
    Opportunity: AI-driven document understanding, summarisation, and structuring can save time and reduce errors.
    Example: Docugami turns complex documents into structured data.

12. Automated Transcription & Translation
    The Need: Audio and video content requires accurate transcription and translation for accessibility and analysis.
    Opportunity: AI tools that deliver fast, precise, multilingual support are in high demand.
    Example: Rev and DeepL provide transcription and translation services.

13. Robotic Process Automation (RPA)
    The Need: Businesses perform repetitive, rule-based tasks that can be automated.
    Opportunity: Integrating AI with RPA enhances efficiency, accuracy, and scalability.
    Example: UiPath helps automate workflows using AI.

14. AI Observability & Debugging
    The Need: Monitoring AI performance in production and understanding model decisions is crucial for trust and compliance.
    Opportunity: Tools that provide explainability and model health tracking are critical for enterprise adoption.
    Example: Fiddler AI offers explainability and monitoring for AI models.

15. Annotation & Labelling Platforms
    The Need: Labelled data is required for supervised learning across multiple data types.
    Opportunity: Crowdsourced or specialised annotation platforms can serve high-volume or niche needs.
    Example: Appen provides crowdsourced annotation and labelling services.

16. AI-Powered Customer Service
    The Need: Companies want to provide rapid, intelligent support without relying solely on human agents.
    Opportunity: AI chatbots and virtual assistants improve response times and learning from interactions.
    Example: Ada powers AI chatbots for customer support.

17. Fraud Detection
    The Need: Financial transactions and networks are vulnerable to fraud and cyber threats.
    Opportunity: AI systems that detect anomalies quickly are highly valuable in banking, insurance, and e-commerce.
    Example: Darktrace uses AI to detect cyber threats and fraud.

18. AI-Driven Market Research
    The Need: Extracting insights from vast amounts of consumer or market data is challenging manually.
    Opportunity: AI can identify trends, sentiment, and opportunities faster than traditional methods.
    Example: Crimson Hexagon provides AI-driven consumer insights.

19. AI-Powered Recruitment
    The Need: Screening and evaluating candidates efficiently is complex and time-consuming.
    Opportunity: AI can streamline candidate assessment, improve hiring decisions, and reduce bias.
    Example: HireVue uses AI to evaluate candidate interviews.

20. AI in Logistics
    The Need: Supply chains are complex, with fluctuating demand, routes, and inventory needs.
    Opportunity: AI can optimise routing, inventory management, and delivery efficiency.
    Example: Llamasoft provides AI-driven supply chain optimisation.

21. AI for Healthcare Operations
    The Need: Hospitals and clinics face administrative bottlenecks and patient care inefficiencies.
    Opportunity: AI can predict patient needs, automate tasks, and improve operational efficiency.
    Example: Olive automates healthcare administrative tasks.

22. AI-Powered Personalisation
    The Need: Consumers expect customised experiences across digital platforms.
    Opportunity: AI-driven personalisation increases engagement, conversion, and retention.
    Example: Dynamic Yield provides AI personalisation for e-commerce.

23. AI for Legal Workflows
    The Need: Law firms and in-house teams handle vast amounts of documents and case data.
    Opportunity: AI can assist with research, contract analysis, and case prediction.
    Example: Casetext provides AI legal research tools.

24. AI in Real Estate & Property Management
    The Need: Property valuation, occupancy management, and market analysis are complex and data-heavy.
    Opportunity: AI can provide predictive insights, streamline management, and optimise listings.
    Example: Reonomy leverages AI for commercial real estate insights.

25. AI for Energy & Sustainability
    The Need: Optimising energy use and predicting equipment failure is critical for efficiency and cost reduction.
    Opportunity: AI solutions can monitor consumption, improve maintenance schedules, and reduce waste.
    Example: Uplight provides AI-based energy management solutions.

26. AI in Manufacturing
    The Need: Manufacturing processes face inefficiencies, defects, and equipment downtime.
    Opportunity: Predictive maintenance, quality control, and process automation via AI reduce costs and improve output.
    Example: Seebo provides AI process optimisation for manufacturers.

27. AI for Cybersecurity
    The Need: Cyber threats evolve rapidly, often outpacing human detection.
    Opportunity: AI can detect, prevent, and respond to attacks faster and more accurately than manual processes.
    Example: Cylance uses AI to prevent cyber attacks.

28. AI-Powered Education Tools
    The Need: Education systems struggle to provide personalised learning at scale.
    Opportunity: Adaptive AI platforms can deliver customised learning experiences for students and professionals.
    Example: Squirrel AI provides AI-driven personalised learning.

29. AI in Finance & Accounting
    The Need: Financial operations involve complex data analysis, auditing, and forecasting.
    Opportunity: AI can automate reconciliations, detect anomalies, and generate predictive insights.
    Example: Kensho provides AI analytics for finance.

30. AI for Governance & Compliance
    The Need: Keeping up with regulatory changes is challenging, especially in financial and highly regulated industries.
    Opportunity: AI solutions can track compliance requirements, flag risks, and automate reporting.
    Example: Compliance.ai helps financial firms stay compliant using AI.

Summary

The AI landscape is moving fast, and the lesson from history is clear: the biggest opportunities often aren’t in striking gold yourself—they’re in giving others the tools to succeed. Early founders, builders, and operators don’t need to be the next OpenAI or Nvidia to make a meaningful impact. There are countless “shovel” opportunities—places where AI is critical, boring, or invisible, but indispensable—waiting to be discovered and built. By focusing on these foundational needs, you position yourself at the center of the AI boom, providing real value and generating sustainable growth.

This is just one of many insights to come. If you found this helpful and want more, you can subscribe to my newsletter at thegrowthframework.com.

by

Glen Smale
Founder, Software Engineer, Techpreneur & Curator of The Growth Framework
A tech media platform uncovering frameworks for growth and success in the real world
www.thegrowthframework.com

Stacking skills lead to success

Each frames it slightly differently:

“Talent stacking means acquiring skills that work well together.”
— Scott Adams, creator of Dilbert

“Spend each day trying to be a little wiser than you were when you woke up.”
— Charlie Munger

“You do not rise to the level of your goals. You fall to the level of your systems.”
— James Clear, Atomic Habits

Alex Hormozi tends to explain it better:

“Skills stack; acquire rare and valuable ones”

In context, this is a powerful yet simple framework for becoming irreplaceable, highly valuable, and ultimately successful in both life and business.

Here’s what it —really means:

• Skills compound. Like interest, when you combine skills, your value multiplies — not adds.

• Rare skills set you apart. Don’t just learn what everyone else knows. Learn what few people know — and even fewer can do well.

• Valuable skills get paid. The market rewards skills that solve painful, expensive problems.

Instead of focusing on just becoming excellent at one thing, Hormozi suggests becoming very good at a few things that work well together — for example:
→ Copywriting + Sales + Marketing + Business Strategy
→ Or: Public Speaking + Leadership + Storytelling + Finance

When these are stacked together, they make you:

• Harder to replace

• Easier to trust

• More in-demand

• And ultimately, more successful

Why this matters:

Because one skill is not enough in a world of high competition.
But a stack of the right skills makes you almost unstoppable.

Here’s to your skills, growth, and success.
The Growth Framework.
— Glen Smale

Introduction: A Personal Confession

Let me be direct.

I’ve lost more time than I care to admit, starting important work only to get distracted, pulled into something else, or convinced that I could “get back to it later.”

That moment rarely came. When it did, the context was gone. Momentum was gone. The result? Another half-finished task. Another day where I looked busy but built nothing meaningful.

Eventually, I learned the hard truth: productivity doesn’t come from dabbling across tasks—it comes from finishing.
And the way you finish is by locking in, starting, and not stopping until it’s done.

That’s The Focus Protocol.

This isn’t a trick, a time management technique, or a trendy habit. It’s a principle that separates those who start things from those who ship results.

Why You’re Always Busy But Rarely Effective

We are surrounded by distractions. Notifications. Meetings. Email. Micro-decisions. Tabs. Messages. The modern work environment rewards responsiveness, not results.

You think you’re productive because you’re juggling multiple tasks and staying “involved.” But the science—and your frustration—says otherwise.

• Gloria Mark’s research shows most people switch tasks every 47 seconds during computer work.

• Stanford research on multitasking found performance drops significantly when people attempt to handle more than one cognitive task at a time.

Every time you switch context, even briefly, you pay a cognitive tax. And that tax compounds throughout the day.

What The Focus Protocol Is (And Isn’t)

The Focus Protocol means the following:

• Identify the one task that truly matters.

• Start it.

• Do not stop until it is done.

This means:

• No breaks.

• No context switching.

• No “checking something quickly.”

• No tab surfing or inbox peeking.

• No timeboxing it and moving on.

• No excuses.

You begin. You finish. That’s it.

This is not about working longer hours. It’s about finishing critical work in one sustained push—without bleeding time and energy into everything else.

Real-World Examples of Singular Focus

Cal Newport

In Deep Work, Newport explains how distraction-free work at your cognitive limit is where your real output lives. He outlines how he writes academic papers, books, and code by eliminating all shallow work until the deep work is complete.

He doesn’t suggest time blocks. He finishes what he starts. That’s the key.

J.K. Rowling

Rowling completed the final Harry Potter novel by checking into a hotel and writing in isolation. No distractions. No partial work. Just complete immersion until the task was done.

Warren Buffett

Buffett is known for ruthless prioritisation. He suggests listing your top 25 goals, then focusing only on the top 5—and actively avoiding the rest until those are complete. He doesn’t recommend multitasking. He recommends deep commitment.

How To Apply The Focus Protocol

This is not a technique. It is a commitment.

Follow these steps.

1. Identify a single, meaningful task
Pick one task that will move the needle. Not urgent tasks. Not shallow tasks. Pick the thing that matters most.

2. Eliminate every possible source of interruption
Turn off your phone. Close your browser. Silence notifications. Let people know you are unavailable. Shut the door. Disconnect.

3. Begin. And don’t stop.
Start the task. Do not stop. If you need to go to the bathroom, do it before you begin. You must remain fully engaged until the task is completed.

4. Finish
Completion is non-negotiable. Even if it’s uncomfortable. Even if it gets hard. You do not walk away until it is done.

5. Review
Once done, review what you completed. Log it. Note how long it took. Note how you feel. You’ve just done what most people can’t: finished meaningful work.

A Real Case Study: Shipping an MVP

A developer I worked with had over a dozen half-built apps. He always said, “I’ll get back to it,” but he never did. We switched his approach.

He picked one project. We agreed on a single rule: start the day with that task and don’t touch anything else until it’s launched.

Six weeks later, he had an MVP, 25 beta users, and real market feedback. He didn’t change tools or learn anything new—he just applied unbroken focus.

How To Protect the Habit

Here are a few ways to reinforce the Focus Protocol:

• Start each day with one key task. Don’t schedule anything else until it’s done.

• Keep a focus log. Write down what you started and what you finished.

• Avoid incomplete transitions. Never stop mid-way through something important unless it’s a true emergency.

• Respect energy and recovery. If the task is big, rest after—not during.

• Don’t stack big tasks. Complete one before moving to another.

Final Thoughts: Why Finishers Win

Dabblers touch everything but complete nothing.

Their hard drives are full of drafts, outlines, and “90% done” files. Their calendars are packed. Their outputs are invisible.

Finishers commit.

They pick a door. Open it. Walk through. And close it behind them.

The world rewards finishers. You don’t need more tools, more systems, or more hacks.

You need to commit fully.

So here’s the challenge:

Pick one task. Start it. Don’t stop. Finish it.
Then do it again tomorrow.

The Focus Protocol isn’t a productivity style.
It’s a discipline.

And it separates those who move the world from those who merely wander through it.

That’s where growth begins.

– Glen

#TheGrowthFramework

Artificial intelligence has been rapidly transforming industries ranging from finance to healthcare, yet public safety and emergency services have remained notably resistant to wholesale disruption. This is not for lack of need but rather due to the extremely high stakes, regulatory complexity, and cultural barriers inherent in these systems.

Amid these challenges, Hyper, a voice AI startup, is carving a path forward by applying sophisticated AI to emergency 911 call centres—delivering not just automation but intelligent augmentation that enhances operational efficiency and improves outcomes for communities.

Understanding the Emergency Dispatch Challenge

The U.S. emergency dispatch system fields over 240 million calls annually, but estimates suggest that more than 50% of these calls are non-emergency in nature—noise complaints, non-urgent medical questions, requests for information, or even prank calls. Despite being low-risk, these calls consume disproportionate dispatcher time and resources, slowing response to genuine emergencies and contributing to staff burnout.

Traditional call centres are ill-equipped to filter this volume efficiently. Manual triage is slow and error-prone, and many legacy dispatch systems lack modern digital infrastructure to support innovative workflows. This creates a systemic bottleneck where human resources are overwhelmed with tasks that AI is well-positioned to handle—provided it can understand and act on nuanced verbal communications in real time.

How Hyper’s AI System Works: Real-Time, Context-Aware Voice Intelligence

At its core, Hyper’s technology is a real-time voice AI platform designed specifically for 911 dispatch environments. Unlike general-purpose voice assistants (e.g., Siri, Alexa) or standard transcription services, Hyper trains its models on an extensive corpus of anonymised 911 call recordings. This unique dataset is a critical competitive advantage, as it enables the AI to understand emergency call dynamics—including emotional cues, urgency levels, local jargon, and call flow patterns.

The AI performs multiple simultaneous tasks:

• Speech recognition: Converts spoken language into highly accurate, real-time text.

• Intent classification: Determines the nature of the call (e.g., non-emergency noise complaint vs. immediate threat).

• Sentiment and urgency analysis: Identifies stress, panic, confusion, or signs of critical danger.

• Task automation: For non-emergencies, it can generate police reports, provide information via text messages, and route calls to the correct non-emergency departments without human involvement.

• Escalation triggers: Automatically transfers complex or ambiguous calls to human dispatchers to ensure no critical situation is missed.

This pipeline enables the AI to act as a trusted first responder assistant, handling routine calls end-to-end and freeing human dispatchers to focus on high-impact, time-sensitive emergencies.

Why Training Data and Domain Expertise Matter

One of the biggest technical hurdles in voice AI is domain adaptation. Emergency calls differ drastically from consumer voice queries. They are unstructured, emotionally charged, and often noisy. Hyper’s access to real 911 audio allows them to fine-tune speech recognition models for these unique acoustic and linguistic patterns. Moreover, their engineers continuously update and improve models using ongoing anonymised call data, creating a virtuous cycle of learning.

This domain specialisation sets Hyper apart from generic AI providers who offer tools that might work well in customer service centres but lack the precision and context awareness required for public safety.

Integrating AI Into a Complex and Regulated Environment

The technical innovation alone would be impressive, but insufficient if the system couldn’t be safely integrated into existing public safety infrastructure.

Hyper has navigated a difficult landscape:

• Regulatory compliance: Emergency dispatch is governed by strict laws regarding privacy, data security, and liability. Hyper’s systems incorporate end-to-end encryption and strict data anonymisation protocols to meet federal and state requirements.

• Legacy systems interoperability: Many dispatch centres use outdated hardware and software. Hyper’s API-first architecture allows integration without massive overhauls, respecting the budget and operational constraints of public agencies.

• Human-centred design: Understanding that dispatchers must trust the system, hyper-designed interfaces that provide transparency into AI decisions and allow dispatchers to easily intervene, override, or audit AI actions.

The Strategic Investors Behind Hyper’s AI Revolution in Emergency Services

Hyper’s successful $6.3 million seed funding round was led by a group of seasoned venture capital firms with a track record of backing transformative technology companies. Each investor brings more than just capital—they bring domain expertise, strategic guidance, and access to networks critical for scaling within complex and regulated markets such as emergency services.

Ripple Ventures

Ripple Ventures is a San Francisco-based venture capital firm known for its focus on early-stage technology companies that are disrupting traditional industries through software innovation. Ripple has invested in numerous startups specialising in AI, enterprise SaaS, and data analytics. Their support signals confidence in Hyper’s ability to harness advanced AI to streamline complex, mission-critical workflows.

Ripple Ventures’ experience helping startups navigate the transition from product-market fit to scale will be invaluable as Hyper expands into new cities and integrates with diverse emergency dispatch infrastructures.

GreatPoint Ventures

Founded by industry veterans with deep operational experience, GreatPoint Ventures invests in early-stage companies in enterprise software, AI, and cybersecurity. GreatPoint’s portfolio emphasises companies solving real-world problems with robust technology platforms.

Their investment in Hyper indicates a belief in the company’s ability to deliver reliable, scalable AI solutions that meet the stringent demands of public safety environments. Moreover, GreatPoint’s operational expertise in navigating enterprise and government sales cycles will help Hyper build partnerships with municipalities and emergency response agencies.

VSC Ventures

VSC Ventures is a lesser-known but highly specialised investor focused on AI, data-driven startups, and infrastructure technology. Their philosophy centres on backing companies that provide foundational technology with a broad societal impact.

VSC Ventures’ involvement reinforces the long-term strategic potential of Hyper’s technology to become an integral part of critical public infrastructure. Their network of technical advisors and AI researchers can support Hyper’s ongoing model development and domain adaptation.

Tusk Venture Partners

Tusk Venture Partners stands out as a venture capital firm with a unique emphasis on startups that engage with government, regulatory, and public sector markets. Founded by Bradley Tusk, who has experience as a political strategist and advisor to technology companies navigating government relations, Tusk Venture Partners is ideally suited to support companies operating at the intersection of tech and public policy.

Tusk’s backing is particularly meaningful for Hyper, which must work closely with municipalities and comply with regulatory frameworks that govern emergency services. Their expertise in public affairs and policy will assist Hyper in managing complex stakeholder ecosystems and accelerating adoption.

K5 Global

K5 Global is a venture capital firm investing in early-stage technology companies with a focus on enterprise SaaS, AI, and security. Known for backing startups that blend innovative technology with strong commercial models, K5 Global supports companies poised for rapid growth in large addressable markets.

K5’s investment reflects confidence in Hyper’s business model and commercial potential within the public safety vertical. Their experience scaling SaaS businesses and building strategic partnerships will be critical as Hyper expands its footprint across the U.S. and potentially internationally.

Why This Group Matters

Taken together, these investors represent a powerful blend of:

• Technical and AI expertise to help refine Hyper’s product and accelerate innovation.

• Operational experience in navigating complex sales cycles in government and enterprise.

• Public policy insight is essential for gaining regulatory approvals and building trust in sensitive public systems.

• Capital resources to support growth, hiring, and market expansion.

Their involvement signals that Hyper is not just another AI startup chasing general voice automation but a company with a mission-driven vision, capable technology, and a credible plan to transform emergency services on a national scale.

For founders and innovators, this lineup underscores the importance of aligning with investors who bring strategic value beyond money—partners who understand the technical challenges, regulatory environment, and market dynamics of your industry.

Looking Forward: The Future of AI in Emergency Response

Hyper is among the first to prove that AI can be deployed responsibly and effectively within emergency services. But this is just the beginning. Future opportunities include:

• Expanding multilingual support to serve increasingly diverse populations.

• Leveraging AI insights to predict demand surges and optimise dispatcher staffing.

• Integrating with smart city infrastructure for more proactive, data-driven emergency management.

• Building broader AI ecosystems that connect emergency medical, fire, and police services seamlessly.

Final Thoughts

For founders and innovators exploring AI’s potential beyond consumer tech and enterprise automation, Hyper’s approach is a compelling blueprint. By focusing on deep domain expertise, human-centric design, and robust integration, they have turned a challenging, critical public problem into an opportunity for AI to genuinely augment human capability and save lives.