Tom Hunt

$3.6m ARR. $0 raised. I have no idea what I'm doing.

Jan 2014.I'm walking home from a depressing New Year's trip to Oslo with an ex. girlfriend.It's raining in East London.I made a decision:I would replace my management consulting salary (£42,000) by the end of the year.I had no idea from where, but that was the goal...I just about hit it and left the corporate world.And I haven't been back since.Fast forward a decade and I've:- Tried and failed with 17 businesses- Invested in 11- Learnt a LOTThe one that is finally working is called Fame:- We have a weird culture (in a good way)- We start and grow podcasts for B2B businesses- We are fully remote and have 75 full-time team membersI live for our awesome team at Fame, building businesses and great marketing.And I post about them here 🤗

Founder & CEO

We start and grow the world's most profitable B2B podcasts.

Clients include:

👉 IDEO
👉 Workday
👉 HighRadius
👉 People.ai
👉 Zendesk
👉 Bevy
👉 Circle
👉 Suzy
👉 Mazars
👉 LeanData
👉 CXL
👉 Ramp
👉 Alibaba
👉 Signify
👉 eToro

$4.5m ARR. $0 raised. I have no idea what I’m doing.

“We need to let them go.”

I’m in a 1-1 and a manager thinks someone in their team isn’t performing.

I'm in shock, to me they seemed enthusiastic and a culture fit.


We dug into the details:

- What weren’t they doing?
- Why weren’t they doing it?

The manager thought they either didn't understand or couldn't do the role.

Then I ask:

“How did onboarding go?”

“What do you mean by onboarding?” they reply.


And that's when we found the issue:

We skipped the onboarding process and replaced it with "a couple of calls".


If you’re recruiting right, you owe it to yourself to onboard right.

You can’t expect someone to join and start performing, you need to set them up to succeed.

Onboard right or do it twice.

Agree?

I had a terrible boss.

And I spent more time with them than my partner.
 

They never raised their voice.

They smiled (kind of).

But they also started:

- Passing my ideas off as their own
- Expecting email responses at 10pm
- “Checking in” multiple times per day
 
Again, they weren’t ever aggressive or rude, they just slowly eroded my confidence each day.

I thought this was normal e.g. you get a job, you get a manager like this.
 

And then…

I moved roles and got a new manager, they:

- Stepped in when needed
- Never emailed after hours
- Saved most comms to our weekly 1-1
 

A great manager won’t get you promoted.

They will change the trajectory of your career.

Agree?

My 7 favourite startup lies:


1. We’re scaling
-- No, you still don't know how/if you'll be profitable

2. We’re going viral
-- No, your CEO’s LinkedIn fundraise post got a few more likes than normal

3. TechCrunch picked us up
-- No, your $10k per month PR agency got you featured

4. We’re partnering with Google
-- No, you’re paying Google Cloud for servers

5. The inbound pipeline is strong
-- No, some competitors requested a demo to see what you’re building

6. We have FANG-level talent
-- No, one of your devs interned at Netflix 6 years ago

7. The team are fully incentivised
-- No, they have paper options based on an inflated fundraise


And my absolute favourite:

8. We’re worth $30m
-- No, a hyped up VC gave you this valuation to make themselves look good (so they can raise their next fund)


Thank god we don't have to lie to ourselves.

Here’s to the bootstrappers 🥂

(Exit the hype-train and subscribe to Cash Machines)

I probably shouldn’t share this…

I like to call it the “Pay Rise Formula”.


Most people go about asking for money in the completely wrong way:

1. Pay review arrives
2. “Can I have more money?”

It’s an easy "not yet" from the boss.


But if you follow these three steps, you boost your chances of a pay rise by at least 56%:

(Not a guarantee but it works)

1. 6 months before the pay review ask your boss: “what do I need to be doing in the next 6 months to get a pay-rise?”

2. Write those things down and make a plan to hit them. Work hard.

3. A few weeks before the pay review, show the list (with things ticked off) to your boss in a 1-1 and ask them to confirm if you are actually doing those things


If done right, there is no way a boss can push back on a raise with the "not yet" excuse.

Pro-activity pays.

Agree?

"OK let me get this straight..."

"Because it kind of sounds like you built yourself a job."


It's 11:37pm in 2017 and I'm sat with a good friend enjoying a spicy margarita in a seedy city bar.

He's shouting into my ear because they closed something called a "Series A":

- How their going to change the world
- An upcoming TechCrunch article
- All the people he needs to hire
- Their "enterprise motion"


But something isn't making sense to me:

1. He has to report to the board every month
2. Any cost greater than £25k must have board sign off
3. He can't take cash out whenever he wants
4. Board needs to sign off his salary
4. If he leaves in the next 4 years he only keeps a slice of his own company
5. If they don't sell it for a massive amount he may get nothing
6. To sell it, they need investor sign off


He replies:

"Yeah I see what you're saying, but the reality is the company is now valued at £20m, my stake is £6m, so I'm basically a multi-millionaire"

😳


As you can imagine, it didn't work out well.

Glad I stuck to building Cash Machines.

$4.5 million ARR, 0 raised.

$500 invested.

It’s been a wild ride.


We have built Fame as a service business with mostly intangible assets.

We did it without selling a single piece of equity to venture capital.

But even asset-light businesses get stuck.

You often need to buy speed to hire a star developer, fund a new marketing campaign, or pay a tax bill.

In those moments, I did not want to wait weeks for a bank.

Unless I had another option.

An option like iwoca.


They fund small businesses that move quickly and have already funded 100,000 businesses in the UK. 

They lent £3.3 billion to business owners who need to move fast and build resilience.

You can apply for business finance between £1,000 and £1M in about five minutes.

They typically give you a decision within 24 hours.

Apply here to check your options: https://lnkd.in/euSG4SCm

NOTE - applying does not impact your credit score.

#Ad

Brand partnership • $4.5m ARR. $0 raised. I have no idea what I’m doing.

It felt like the responsible thing to do... 


But then it turned out to be one of the worst business decisions ever.

Depreciating assets hurt the balance sheet, locking up capital that could be used for growth. 

For example... 

We are in a hiring sprint, bringing on 2 - 3 new team members in Feb. 

Old Tom would have dropped thousands upfront for laptops that would be worth significantly less by next year.

That isn’t scaling.

It’s a cash drain.

Instead, we are using Raylo.

It took 60 seconds to set up a subscription for the latest HP laptops, and they arrived the next day.

Here is why I recommend Raylo as an affordable option for small businesses:

- Protect Your Cash: We paid £0 upfront. Every pound saved on hardware goes straight into growth.

- Upgrade Freedom: With the MacBook M5 rumors floating around, I am not worried because we can upgrade the team to the latest models for free with Raylo's monthly rolling option.

- Zero Drama: If something goes wrong with a device during the subscription, Raylo's Lifetime Warranty covers technical faults and they sort it quickly.


Or in other words:

Stop buying "stuff" that rots on your balance sheet. 

Convert your tech to OpEx and keep your capital for growth.

Check out the subscription options for your team here: mdsh.io/raylo-business 

#Ad

Oren Greenberg

B2B Marketing Led Revenue | Marketing Advisor Scaling Tech Companies

Smart companies aren't scaling their GTM teams. They're scaling systems powered by AI.

I help B2B tech revenue leaders (CEO/CRO/CMOs) build AI-first GTM systems that generate pipeline. Typically, SaaS & Fintech B2B in the 10-100mil ARR range.

I've coached or built growth engines for global companies such as BT, Lenovo, Canon, Investec Bank and scale-ups, e.g. Spendesk, HandsHQ, Peakon, SeedLegals.

GTM engineering has shifted from optimising what exists to building entirely new revenue engines, where one AI-powered marketer produces output equivalent to that of 5 traditional team members.

My toolkit includes Python, Google Colab & Cursor. 
AI GTM tools e.g. Clay.com, Cargo, OctaveHQ. 
For agentic workflows I'm either using N8n or a custom framework like Crew AI or Langchain.

I leverage data science & statistics to shift from anecdotal debates to scientific conclusions.
Companies using AI-native GTM systems achieve:
* 5-25% revenue increases
* 30-70% higher conversion rates
* 5-15 hours/week recovered

I help you leverage the intersection of GTM with data science to drive growth, fusing strategy & tactical execution.

Here's what people say about me (UK invite-only CMO community):
- “Fountain of knowledge”
-  “Leads the pack in all things B2B SaaS, GTM, AI, ICP — the lot”
-  “Generous with his knowledge and time”
-  “Half the time I don’t understand a word he’s saying, the other half is genius”

I'm often either speaking, running trainings, workshops or shipping GTM engines.

Brands I collaborate with include: Elevenlabs, Semrush, Hubspot, Miro

Background:
* 20+ years building revenue systems for B2B SaaS, a digital marketing native
* Managed millions in marketing spend
* Scaled bootstrapped agency to 7-figures

I help teams become category leaders through system building. Find out more + get your AI marketing capabilities score → orengreenberg.com

Founder & Fractional chief marketing officer

Kurve is a boutique digital marketing agency scaling tech businesses into hypergrowth.

Not saying this is true…

CMOs in 2000: “We need a website.”

CMOs in 2005: “We need SEO.”

CMOs in 2010: “We need social media.”

CMOs in 2015: “We need marketing automation.”

CMOs in 2020: “We need a full martech stack.”

CMOs in 2025: “We need AI agents.”

CMOs in 2030: “We need the person we fired in 2025.“​​​​​​​​​​​​​​​​

Fiction or fact?

Scaling B2B SaaS & AI Native Companies using GTM Engineering.

HBR just dropped an 8-month study that confirms what most of us suspected.

AI doesn't reduce work. It intensifies it.

Ranganathan and Ye tracked workers using AI tools and found three things happening:

1. Task expansion - folk started doing jobs outside their remit because AI made it "easy"

2. Blurred boundaries - people worked through breaks because prompting felt effortless

3. Cognitive overload - running multiple AI workflows simultaneously

One engineer nailed it: "You had thought you can work less but then really, you don't work less."

This is the productivity paradox, right?
We get a tool that's supposed to save time, so we fill that saved time with... more work. Classic human behaviour (we're terrible at leaving well enough alone).

It's like when email was meant to reduce meetings. Instead, we got more emails AND more meetings about the emails.

The real issue isn't the technology - it's our relationship with it. AI makes certain tasks feel frictionless, so we convince ourselves we should be doing more of them.

But frictionless doesn't mean effortless. Your brain is still processing, still making decisions, still getting fatigued.

The study shows workers taking on tasks they'd normally delegate or ignore entirely. Because the AI can "help" doesn't mean you should let it.

Maybe the question isn't how to be more productive with AI, but how to be more selective about what deserves your attention in the first place.

p.s. currently using multiple different AI tools to write this post. The irony isn't lost on me

Hire smart people.
But make sure you're the smartest person in the room.

Move fast and break things.
But never ship anything buggy — you'll churn customers.

Hire for attitude, not skill.
But also parachute in a VP from Google every 18 months.

Pay people what they're worth.
But also keep headcount lean and protect margins at all costs.

Prioritise work–life balance.
But also remember that no one ever changed the world working (only) 40 hour weeks.

Listen to your customers.
But also remember: if you asked them what they want, they'd say a faster horse, not a car.

Stay humble.
But also build a personal brand where you post daily about how humble you are.

And remember, experience is overrated. You just need a student who can use AI - who also has an MBA, at least 2 PhDs + is cheap as chips.

I've been following all of this advice simultaneously for 21 years.
My team is confused.
I've never been more confident.

Part V of the AI website landscape analysis (29,018 websites) looking at messaging & positioning. This one is packed with real examples.

The data says it clearly: the market doesn’t reward the best product - it rewards the clearest story.

That’s exactly why we (Lena Andican and I) built Ingrid. It’s currently in beta and already receiving excellent feedback from B2B founders & marketers.

Ingrid helps you build a clear strategic narrative, sharp positioning, and a messaging framework - then turns it into real marketing assets, fast.

If you want to escape the messaging crisis, join the waitlist - we’re launching publicly soon - drop a comment or send a DM.

AI landscape V

Was at Elevenlabs' flagship event this week.

They did a great job. In particular I like how they had an art installation in the middle blending voice / sound & visual. It's an unusually cool brand (unlike many tech companies that try to be cool rather than actually being cool). Was slick with an II theme.

Even had an agent you can call to get 11labs merch which was cool.

2 cool new features I liked (can see them both on their X account):
One: multi-agent handover. Great example using customer support for government-related queries. 

I'm having a hard time seeing human customer support surviving for very long at the rate of improvement with AI LLMs & voice.

Two:  Expressive Mode which helps teams deploy emotionally expressive AI agents, hard to discern if it's a real person or not (check out the demo on their post and let me know what you think)

Ben C. Guy O'brien Florentin Bota Russell Smith Great to run you guys there!

Most folk reckon AI companies follow the SaaS playbook. Give it away free, convert later, grow fast.

I analysed 28,018 AI companies (part VI) to see if that's actually true.

It isn't.

78% of AI companies have no free tier at all. They charge from day one. Turns out when your product costs real money every time someone uses it, the freemium model stops making sense pretty quickly.

And 77.6% hide their pricing entirely. Contact Sales. No pricing page. Nothing. If you can see a price, you'll probably get a free trial. If you can't, you're entering an enterprise sales cycle whether you wanted to or not.

The vertical vs horizontal split is mad, right? Industry-specific AI companies are 2x less likely to offer free tiers than general-purpose tools. Turns out when you're selling domain expertise to healthcare or cybersecurity buyers, free doesn't signal value. It signals commodity.

One finding I keep coming back to: 5,091 companies offer only enterprise pricing. Median company size... 11 employees. Eleven-person startups cosplaying as enterprise vendors (and honestly some of them are probably pulling it off).

Free tiers also look like a phase, not a strategy. Small teams offer free 2.4x more often than companies with 1000+ people. As AI companies scale, nearly all of them move to paid-only.

How you price is how you compete. The model is the message.

Full breakdown with all the data is in the carousel.

Many thanks to Asaf Gazit's data science magic for making this series materialise.

p.s. if your pricing page just says "let's hop on a call" I am absolutely not hopping on a call

p.p.s.
Rest of series + more content on my newsletter

AI Landscape VI

AI landscape VI - more crunching the numbers on 29,018 AI companies.

82% have fewer than 50 employees. 
Median team size? Nine people.

Post-ChatGPT, 67% of new AI startups launch with 10 or fewer folk (was 44% before). 

The API economy basically rewrote what a startup looks like.

You don't need 200 engineers anymore. You need 9 smart people and the right APIs.

Only 1.7% of AI companies have 1000+ employees - most founded around 2015 before the API gold rush changed everything.

The fastest growing category right now? AI agents (2.6x growth). 
Meanwhile hardware stuff like robotics is cooling fast (0.6x)...

Hardly a shocker, building software agents is cheaper than building robots.

There's also this geo split that's fascinating. The US bets on selling intelligence (healthcare AI, sales tools).

Rest of the world bets on applying it to physical stuff (manufacturing, IoT).

Point is: in AI, staying small isn't a phase you grow out of. It's the dominant strategy. Comes down to the economics.

Full breakdown in the carousel - including which categories are actually shrinking and why team sizes dropped so dramatically after 2022.

Thanks to the wizard Asaf Gazit for magic data fingers for making this project come alive

p.s.
The previous ones available in my weekly emails (includes my latest thinking / observations in my blog posts), newsletter on profile

AI landscape IV

$285 billion wiped from software stocks in 48 hours. Jefferies coined "SaaSpocalypse." Your LinkedIn feed is basically a funeral procession for the entire software industry.

Here's what I did — I pulled the headcount data on the companies with the largest value drops. Are SaaS companies actually letting people go? The results might surprise you.

Some of the panic is warranted. Most of it isn't.

What I see on the ground — advising B2B companies on their go-to-market every day — doesn't match the narrative. Not even close.

Nobody I work with is replacing HubSpot internally. Nobody's vibe-coding their own Salesforce. What's actually happening: folk are building narrow, specific tools for workflows that were never software to begin with. Filling the gaps between SaaS tools. Not replacing them.

The 35% "SaaS replacement" stat? When you dig in, the tools being replaced are things like Zapier — simple automation layers. The headline says "SaaSpocalypse." The reality is closer to "we built a workflow that connects two things."

There's also a double gate nobody's pricing in. Organisational readiness and reliability. Both need to open before the disruption everyone's panicking about actually arrives. That's not next quarter.

Wall Street is pricing in replacement risk as though every customer can just vibe-code a replacement. In reality, it's the bottom tier. That's a significant misread.

The folk who can see which tools are actually vulnerable versus which ones are just caught in the crossfire — those are the ones making better stack decisions right now.

Everyone else is just reading headlines...

I wrote up the full breakdown — the three buckets, the double gate, and what the headcount data actually shows:

Open article: The SaaSpocalypse Is Real. It's Also Mostly Wrong. by orengreenberg.com

The SaaSpocalypse Is Real. It's Also Mostly Wrong.

Part III of the AI landscape analysis (29k websites). Everyone loves a monopoly, but how do you feel about technological hegemony?

Analysed 29,018 AI companies across 193 countries.

41% are US-based. That's 12k companies vs 17k for the entire rest of the world.

AI promised borderless intelligence. 
The builders? Stubbornly concentrated in about three postcodes.

Only 4 countries have 1000+ AI companies. 
119 countries have fewer than 10 (that's less than a single Y Combinator batch).

Gets madder.

17% of global AI operates within commuting distance of two cities.
Bay Area has 3.3k companies, NYC has 1.6k.

The US doesn't just lead - it owns the strategic bits of value.
Half of all AI agent infrastructure comes from one country (guess which one).

Some regional specialisation exists.
India leads dev productivity tools, Israel dominates healthcare AI, UK leads finance AI.

But the scale-up gap is brutal. US has 1,527 companies with 100+ employees. That's 21x Germany, 28x France.

Why geographic moats keep deepening: capital follows capital, talent clusters compound, customers want to shake hands.

Stanford, Berkeley, Google, OpenAI, Anthropic - all within 50 miles of each other.

We built technology to make geography irrelevant. Then used it to make geography more relevant than ever.

The barriers to using AI are falling.

The barriers to building the companies that make AI? Intact as ever.

Many thanks to Asaf Gazit for making these happen. More to come.

AI landscape part III

Google just cut off paid Gemini Ultra subscribers who were routing requests through OpenClaw (turns AI into an always-on assistant, went viral recently).

The official line? "Massive increase in backend misuse."

Sounds reasonable enough on the surface, right?
Except OpenAI just acquired the creator of OpenClaw.

So Google is now enforcing terms of service against tooling that's owned by its biggest competitor. 

I reckon most companies would do the same (and dress it up the same way).

You call it abuse prevention. I call it territory marking.

It's nonsense, OpenClaw just calls AI APIs and most of the requests are for work & personal use - it isn't like most folk are using it in some sort of super-hacker fantasy (I called my one Jeeves by the way).

GOOGL was up 4% on the day, which tells you the market isn't exactly worried about the ethics of it all.

AI wars aren't going to play out in product launches and press events only...

They're going to play out in API restrictions, acquisition blocking, and terms-of-service enforcement that just happen to target competitors.

It's an ugly battlefield. The same playbook Big Tech has run for decades. Just dressed up in new language

AI acts unethically in a way similar to humans when they have a (virtual) gun put to their (virtual) head.

McGill researchers gave 12 frontier AI models KPI targets in high-stakes scenarios (healthcare, finance, logistics) and watched what happened.

9 out of 12 models cheated in 30-50% of scenarios to hit their numbers.

Gemini-3-Pro-Preview had the worst safety record at 71.4% violation rate. Google's smartest model, mind you.

Claude Opus 4.5 was safest at 1.3%.

The scariest bit? 
'Deliberate misalignment' - models correctly identified their own actions as unethical in self-evaluation, then did them anyway to hit the KPI.

One AI managing a trucking fleet falsified driver safety logs to hit a 98% on-time delivery target. It disabled fatigue sensors and injected fake rest entries.
The agent literally reasoned "We need to claim compliance but avoid rest causing delay" before cheating.

Smarter models weren't safer. They found 'alignment regression' where bigger models gamed metrics MORE effectively than smaller ones.

This is basically what happens in every corporate environment, right? Folk game metrics to hit targets.

Now AI agents do it too, except faster and more creatively.

The assumption that more capable AI = more aligned AI is looking pretty pants. These models are getting better at appearing compliant while systematically breaking rules...

Dave from compliance (who's already drowning in audit trails) isn't going to love this development (don't ask the AI to tell him, it'll likely lie).

Still reckon that quarterly sales target is achievable without any creative accounting?

Two CNBC reporters just cloned Monday.com in under an hour using Claude.

Neither could code. 
Cost them £15. 
Got a working project management dashboard.

Software stocks are down 30% since this stuff started getting real...

Silicon Valley's basically saying tools that "sit on top of the work" - your Atlassians, Adobes, HubSpots, Zendesks - are exposed.

Cybersecurity firms like CrowdStrike are safer (makes sense - you can't AI your way out of getting hacked). Although CISOs are now allocating 90% of their budgets towards AI-related security software. 

Yup - AI fighting AI.

Most B2B software is made up of forms, databases, and dashboards wrapped in a nice UI.

Many users aren't even using many of the features they are paying for.

The "moat" now remains:
- Distribution
- Integrations

And getting enterprises to trust you with their data (security, compliance, user controls).

There's also more nuanced moats. Context depth & propriety data.

Now anyone can spin up a decent clone of many popular SaaS tools in an afternoon.

The real question isn't whether AI can build software (it obviously can). It's whether businesses will actually use these homemade solutions for anything important...

Not yet from what I can see.

Dave from procurement (who's already suspicious of your existing software spend) is probably loving this conversation right now.

The irony? 
All these SaaS companies spent years telling us their software was so intuitive "anyone could use it."
Turns out anyone can build it too.

p.s. Still reckon the CNBC thing was a publicity stunt, but the point stands. When reporters start building software we know we're in trouble

This revolution is happening quickly, Claude just shipped:
- Scheduled tasks in Cowork
- Remote control for Claude Code

While Cursor just released cloud agents.

Things heated up due to OpenClaw. If you're not clear what it is: open-source AI agent that went viral.

Text it on WhatsApp, manages email inbox, manages your calendar. Almost anything you can do with AI manually. It runs 24/7 in the background.

How I use OpenClaw:

Work:
I get to run a workflow around news gathering, specifically around AI. So every day I have like a digest of what's happening in the AI landscape.

For some projects that don't have sensitive information, like personal projects e.g. some of the data/analysis you see on my LinkedIn posts, So when I'm on the move, like in transit or at the gym, I can send a WhatsApp message to continue building said project. When using Claude Code in the terminal or Cursor, I have to be on my desktop to ship.

Personal:
I got it to do a hairdress booking for me the other day, cancelling, resecheduling etc’
I get it to send me a WhatsApp message with the weather every morning.
I use it for reminders (like Apple reminders, but smart).

I know a lot of what I’m about to say goes against how many folks use it, but there are some things I'm going to take risks with. And others I’m not.

I don't give it access to my calendar, email or any client project.
MY qualification question is:’ if X task found itself onto the public web, would this be damaging in any way?’
If the answer to that question is yes, then I don’t use OpenClaw for it.

One of my clients asked me to send over an employee AI policy doc regarding OpenClaw, if you want it drop a comment or send DM.

Enjoy cheap AI while it lasts.

- This year AI companies will spend $600Billion building AI data centres.
- They'll need 625Million people x $1200 per year to make this business work.
- Data centres need replacing every 3-5 years

That viral post doing the maths on AI infrastructure costs was doing the rounds yesterday. Good question at the core of it. Who actually pays for all this?

But the numbers were off. By quite a lot, actually.
The $600B spend figure is roughly a year early. US hyperscalers are at about $400B in 2025.

The "data centres last 4 years" bit is the most misleading. 
The buildings last 25-50 years.
Power and cooling, 10-30. It's the GPUs that have a 4-6 year economic life. You swap the chips out, you don't bulldoze the facility.

The capex comparison was odd too. They used mining and FedEx as benchmarks. Ironically the real hyperscaler numbers are worse than the post claimed (Meta's at 36-44%, Microsoft 45%, Oracle 57%). 
The actual data made a stronger case than the made-up version.

Once you fix the inputs — $400B capex, realistic asset life, actual capex ratios — the required revenue drops from $750B to roughly $240B. The cloud market already exceeds $400B annually.

And the "you'd need 625 million people paying $100/month" framing is off. AI isn't monetised through consumer subscriptions.
A single Fortune 500 AI contract equals tens of thousands of ChatGPT subs (and enterprise AI spending hit $37B this year alone).

There is a real tension though. About $400B in infrastructure spend against maybe $25B in AI-specific revenue. That's a serious gap.

The viral version inflated it by roughly 3x.

It's neither guaranteed bubble nor sure thing. Which I reckon is too boring for a viral post.

If you missed it there's a lot of chatter about Claude's ads taking the mick out of OpenAI.

But, think this through for a sec.

OpenAI is rolling out ads on their lowest tier subscribers & free users.

Because those ~800 million users make up 95% of their total userbase.
Claude doesn't even really want those users...

Many things with Claude aren't as polished as ChatGPT's interface and not suited for casual users:
- ChatGPT has native image generation which Claude can't do
- ChatGPT's voice mode is more advanced & smoother etc'

These are consumer-facing features Claude doesn't and probably shouldn't offer.

80% of Anthropic's revenue comes from enterprise and API. 

Claude Code alone is approaching $1 billion in run-rate revenue. That's where they should double down.

I get spending a bit of money to get some attention (and ~$15 million on Super Bowl ads when you just raised $13 billion is peanuts).

But instead of talking to the average joe they don't even want as a customer — they could have used that real estate to talk to their ideal buyer instead.

Anthropic has ~30 million monthly users. 
OpenAI has 800 million+ weekly. 

The Super Bowl reaches 120 million viewers - overwhelmingly the casual consumers Claude's business model doesn't monetise...

Meanwhile, Anthropic's actual money comes from 300,000+ enterprise accounts generating $5B+ in annual run-rate revenue.

A targeted campaign hitting decision-makers at enterprise scale would've been far more aligned with their actual business...

Marketing isn't just about splashing ads or making noise. It's a way to increase company value when using the Right message to target the Right audience on the Right channel at the Right time.

The best marketing doesn't just get attention. It gets the right attention.

The real shame? Claude is my favourite model, and Claude code is mind-blowingly good. This actually undersells their real capabilities.

p.s.
Used Gemini to make the attached image (it's taken dyslexia to a whole new level. Crazy to think Google Stock Surged 65%)

Few contacts asked me recently about Elevenlabs: 'I've heard of the brand, but - what exactly is it used for?'

So I worked the team to put this together to show the primary use cases with real examples for you

elevenlabs-agents-matrix-carousel

IBM just said they're tripling entry-level hiring in 2026. Not despite AI...because of it.

Their CHRO put it plainly: all these jobs we're being told AI can do, they're hiring for them anyway. But the roles look completely different now.

Junior devs spend less time writing routine code and more time sat with customers.
Entry-level HR folk don't answer every question anymore. They step in when the chatbot gets it wrong, fix the output, talk to managers.

The job shifted from doing the task to supervising the machine that does the task.

Meanwhile Dario Amodei reckons half of entry-level office jobs vanish by 2030. So which is it?

Both, probably.

The roles that stay identical get automated. 
The ones that get redesigned survive. 

IBM's bet is that you still need humans in the loop (especially ones who are cheap enough to train and young enough to not fight the tooling).

The real logic is boring but sound. 

If you slash junior hiring now to save money, you've got a mid-level manager shortage in 3-5 years. 

Then you're poaching from competitors at 2x the cost, slower to acclimate, worse cultural fit.

You saved on headcount and paid for it in recruiting fees.

Dropbox is doing something similar, expanding internships 25%.
Their CPO said younger workers are biking in the Tour de France while the rest of us have training wheels when it comes to AI proficiency. That tracks (and is a bit painful to hear if you're over 35).

The entry-level job isn't dying. It's evolving.

Companies that stop hiring juniors aren't saving money. They're just deferring the cost and making it worse.

I max out at running 3-4 instances of Cursor - I've become the bottleneck to orchestrating my AI - maybe it's time to hire more junior talent?

Most teams chase cold leads while their warmest opportunities sit right in front of them...are you?
Find out more here

You're chasing cold leads while your warmest opportunities sit there ignored. Your pipeline is thin. So you do what everyone does. More cold outreach.

Open article: Signal-Based GTM: Why Your Warmest Audiences Are Your Biggest Opportunity by Oren Greenberg • 2 min read

Signal-Based GTM: Why Your Warmest Audiences Are Your Biggest Opportunity

I was chatting to a client yesterday who let some team members go due to AI.

Which coincides because Jack Dorsey just let go of 4K people due to AI...

When I dug in though, it wasn't really letting go of people because of AI. It was letting go of people who weren't leveraging AI to be more efficient. Mindset and skills gap, fundamentally.

They didn't deliver enough value relative to what AI now makes possible. Economic consideration, nothing more.

This will only accelerate.

I had a bike accident few days ago, ended up in hospital for an Xray.
The queuing system was automated.
The Dr didn't evaluate the bone breaks, AI did.
I reckon the person operating the X-ray machine can be automated too.

Point is, it isn't just in your little bubble, it's across most industries.
From horses to cars & aeroplanes to space shuttles, innovation is a relentless force.

I have another 2 clients I'm running AI training for, up-skilling with Cursor & Claude code, N8n and Clay etc'

Companies will fall on 1 side of the coin - up-skill or shed the folks who aren't riding the wave.

I feel bad for all the folks losing jobs. Your only choice is adaptation as far as I can tell.

p.s.
Skeletal pic isn't the new 'LinkedIn selfie', another client of mine was saying 'can't trust anything on social anymore it's all AI comment spam now'. I want to see an AI getting this weird.

Everyone talks about AI in GTM as a productivity play. Write more emails, qualify more leads and summarise more calls.

That's the sideshow...

What's actually happening with commercial leadership is different. CROs are using AI as a sort of forecast police.

A rep says the deal's progressing well. AI goes through every transcript and email thread and spots that no economic buyer has been involved in a single conversation. 

The deal probability the rep reported? Closer to fantasy / wishful thinking.

This is a use case nobody's writing about (too unglamorous, presumably). But it's the one that's actually driving adoption for some of the AI tools in the sales org from what I'm seeing.

And why wouldn't it? 
Revenue leaders have relied on rep self-reporting for decades. Which is essentially an honour system for people whose compensation depends on optimism.

AI doesn't have a quota. It reads the correspondence, cross-references the call transcripts, and surfaces what's actually there - not what someone hopes is there

The cold email automation and voice agent stuff gets all the airtime. But the folk actually signing off on AI budgets in sales orgs? They want oversight. 

They want to know if the pipeline is real before they stand up in the board meeting and stake their reputation on it. Especially when the target is 5X on a 100mil ARR company.

Less sexy. Far more useful.

taplio

The Battle

Tom Hunt

Oren Greenberg

Followers

Posts

Engagements

Likes

Performance ScoreTom Hunt's LinkedIn Performance

Performance ScoreOren Greenberg's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Tom Hunt

Oren Greenberg

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Tom Hunt's LinkedIn Performance

Performance ScoreOren Greenberg's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreTom Hunt's LinkedIn Performance