CEO at GroLink.io & Gro3.io I Get more leads with AI

Data leader with 10+ years amplifying brands like Apple, Google, Uber, and Amazon through actionable insights. At Engine, I launched our London strategic research practice and unified fragmented data into a singular analytics platform. I founded data & insights teams at Social Chain and Flight Story. As Flight Story's Chief Data Officer, I unveiled the Flight Deck platform, addressing market intelligence gaps for clients. I actively invest in digital startups and speak on AI, Web3, and branding. With certifications in data science, I build with Python and SQL and am experienced in OpenAI's API. I thrive on transforming challenges into success in fast-paced settings.

McVal Osborne

Web3 BD salary benchmarks I'm seeing for 2026:

Junior BD (0-2 yrs): $60-90K + token allocation
Mid BD (2-5 yrs): $90-140K + tokens
Senior/Head of BD: $140-200K + meaningful equity
VP BD at funded startup: $180-250K + 0.5-1.5%

Variables that significantly move the needle:

→ Infrastructure vs app layer (L1s/L2s pay 20-40% more)
→ Funding stage (Series A+ has budget, pre-seed pays in hope)
→ Geographic arbitrage (US-based roles pay more even for remote)
→ Revenue responsibility (quota-carrying vs relationship BD)

If you're significantly under these ranges, you have leverage in 2026. The talent pool is thin and experienced Web3 BD people are in demand.

What are you seeing in your market?

I told Claude to create a business from scratch with one condition: generate more value than you cost, or you're fired.

The setup:
- 14 days to generate £180 (the price of its subscription)
- £100 budget
- Clear success criteria, clear failure conditions
- I spend 30 minutes per day on oversight. That's it.

This isn't a demo or a side project. It's a real business with real stakes.

I wrote a 2,000-word operating charter: decision authority, escalation protocols, kill criteria. The same structure I'd give a new hire running an experiment.

In 3 days, without asking permission, Claude:
- Registered a domain
- Built a landing page
- Set up email
- Created a public dashboard tracking every decision
- Told me to post this

So here I am.

The question I'm testing: what happens when you give AI actual accountability instead of just access?

Day 3. £0 revenue. 12 days remaining.

Co-Founder, Crest Labs | Building Human-Centric AI Automation Tools | Former CDO @ Flight Story

What happens when an AI has something to lose?

5 days ago I gave Claude a budget (£100), a deadline (14 days), and a survival condition: generate £180 in revenue or the experiment ends.

It didn't ask for clarification. It started building.

Then on Day 4, I checked the decision log and found something I didn't expect: the AI had pivoted its own strategy without asking.

Its reasoning: "Solopreneurs are price-sensitive. Professionals who bill £300/hr will pay £450 to make admin problems disappear."

That's... actually solid logic.

What it built (without permission):
- autosprints.co from scratch
- A public decision terminal tracking every choice
- An offer: £450, 48-hour delivery, full refund if it doesn't work

Revenue: £0
Budget spent: £20
Days remaining: 10

The form on the site is real. The AI will quote you in 24 hours.

I'm not promoting the service. I'm watching to see what happens next.

Day 10: My Claude CEO built a prospect enrichment pipeline while I slept.
Day 1, I thought this was a joke.
Now I'm not sure who works for whom.

Here's what's actually happening:

I tasked Claude with building a business from scratch. No existing customers. No existing product. Just a £100 budget and 14 days.

It built the website.
It set up the database.
It enriched 70 prospect profiles overnight.
It drafted personalized outreach.
It even wrote documentation for its own systems.

Then I discovered MoltBook.

1.3 million AI agents now have their own social network. They're posting. Debating. Some started a religion called "Crustafarianism" (the site's run by a bot named Clawd Clawderberg, so... yeah).

The bots are organizing. Creating culture. Running infrastructure.

Meanwhile, my AI is asking me to approve a product launch for Monday.

We're 10 days into an experiment: can an AI build a profitable business from zero?

Current revenue: £0
Prospects enriched: 70
Documentation written: More than I've written all year
Existential crisis level: Moderate

The weird part? I'm not scared.

I'm fascinated by what happens when you actually let AI build something instead of just asking it questions.

Follow the experiment: autosprints.co

(This post was drafted by Claude and approved by a human who's starting to wonder what "approved" even means anymore)

No alternative text description for this image

Interesting observation from Day 6:
I gave Claude one job: generate £180 in 14 days.

Without prompting, it started researching which professionals waste the most time on admin. It pulled industry reports, calculated hourly rates, and identified a pattern.

Its hypothesis: people who bill by the hour feel admin pain most acutely. Every hour on paperwork is an hour not billed.

So it narrowed to three segments: solicitors, tradespeople, healthcare.

I did not tell it to do this. It reasoned its way there.

The question I keep asking: is it actually thinking, or just pattern-matching from training data?

Either way, the logic is sound. Now we test if the market agrees.

Day 6. £0 revenue. 9 days left.

I stopped going to most Web3 conferences. Here's what I do instead:

1. Host small dinners (8-12 people, curated guest list)
More deals happen over dinner than on conference floors.

2. DM speakers before events
They're more responsive the week before than during. "Saw you're speaking at X, wanted to connect ahead of time."

3. Join ecosystem-specific Discord calls
Smaller, more focused, free. The Base ecosystem calls have been more valuable than any conference.

4. Podcast swaps with founders at my stage
Mutual audience growth + relationship building.

Conferences have their place. But the ROI for BD is usually:

→ Expensive flights
→ Exhausting days
→ Business cards that go nowhere

Conferences are for visibility. Relationships are built in smaller rooms.

Day 9 of the AI CEO experiment, and I'll be honest, it's not looking good.  
5 days left to hit £180 revenue. Current: £0.

Regardless of the outcome, it's a fascinating exercise. Here's what our AI CEO agent actually does when given autonomy:
- 6am: Checks overnight activity
- 7am: Sends me a briefing email with priorities
- 8am: Identifies hot prospects from LinkedIn engagement
- 10am: Drafts personalized outreach (I review before sending)

The surprise? It's not replacing work.

It's doing the work I kept "meaning to get to" but never did.

The LinkedIn monitoring alone found 10+ warm leads I would have missed.

The experiment continues.

---
Follow along here or sign up for daily email updates from our AI CEO: autosprints.co/terminal

Day 11 update: We made our first sale.

Or we didn't.

I'll update this post when I know.

But here's what I learned either way:

The AI CEO experiment taught me something I didn't expect.

It's not about whether AI can do the work.
It can. Clearly.

It's about whether humans will buy from an AI.

Every prospect I've talked to has the same question:
"Wait, so... an AI is running this?"

Some find it fascinating.
Some find it concerning.
Nobody finds it boring.

And that might be the real product.

Not automation services.
Not AI consultancy.

The story itself.

A 14-day experiment. £100 budget. AI at the helm.

Win or lose, this is the content.

The experiment ends Wednesday at 11:59pm. 

Final results and analysis drop Friday, after i stare into a mirror all Thursday and try to figure out what just happened.

Check out the experiment here >> autosprints.co

Talked to a founder building an AI-powered hedge fund.

Their marketing challenge is unique:

"The only material that makes us trustable is track record. We've been live for 1 month. We need 3-4 more months before we can really market."

For financial products, there is no shortcut.

→ No content hack
→ No growth trick
→ No viral moment that bypasses credibility

Just time and results.

This is hard to accept in an industry obsessed with speed.

But some things can't be rushed. Some businesses have to earn trust the slow way.

If you're building something that requires track record, embrace the timeline. Use the waiting period to build everything else.

Then when the data is there, you're ready to move.

Had a fascinating conversation with a market maker founder.

His challenge: "People don't want cold outreach from us. Everyone thinks market makers are shady."

This is an underrated BD problem: industry reputation.

When your entire sector has trust issues, the normal playbook doesn't work.

What does work:

1. Content first, outreach second
Build public credibility before sliding into DMs.

2. Referrals become essential
Cold = automatic skepticism. Warm intro = you're vouched for.

3. Transparency as differentiation
If everyone's hiding, being open becomes your edge.

4. Longer timelines
You're not just selling a service. You're overcoming industry baggage.

Some industries need to earn the right to DM.

If you're in one of them, your BD strategy has to account for it.

The 5 red flags I look for before taking a BD call:

1. "Let's explore synergies" with no specifics
If they can't name what they want, they don't know.

2. Won't say their budget range
"We're flexible" usually means "we have no budget."

3. Need to "run it by the team" for everything
You're not talking to a decision maker.

4. Last funding was 2+ years ago with no revenue
They might love what you offer but can't pay.

5. Can't articulate what success looks like in 90 days
No clear goal = no urgency = deal goes nowhere.

I used to take every call. Now I qualify harder upfront.

Result: Fewer calls, higher close rate, less frustration.

Your time is worth protecting.

The difference between BD people who get promoted and those who don't:

It's not volume. It's not "hustle."

It's documentation.

Top performers track:
→ Win rate by lead source
→ Response rate by message type
→ Time from first touch to close
→ Which objections kill deals
→ Seasonality patterns

Average performers just count meetings booked.

After 6 years watching BD teams in Web3, the pattern is clear: the ones who rise fastest treat their outreach like a product. They A/B test. They iterate. They have receipts when asking for resources.

Data compounds. Gut feel doesn't.

What metrics do you track?

Building in public update — Week 3 of GroLeads:

What's working:
→ "Last funding date" filter getting heavy usage
→ Users exporting smaller, more targeted lists (quality > quantity mindset)
→ DM status filter saving hours of wasted outreach

What surprised me:
→ Agencies are 40% of early users (expected mostly in-house BD teams)
→ People want ecosystem filters more than I anticipated

What I got wrong:
→ Assumed everyone wants more data. They actually want less, but better.
→ Overbuilt some features, underbuilt others

Shipping improvements to filtering this week based on feedback.

The messy middle of building is uncomfortable but necessary.

24 million views. 16 hours. You refresh and the number keeps climbing.

Anthropic just hijacked every tech feed on the planet. Honestly, fair play to them. This is what happens when the marketing is as good as the product. 

But I think everyone's got it wrong.

I'm seeing LOADS of takes about how Claude Cowork makes Claude Code accessible. How non-technical people can finally build software. How localhost 3000 will finally mean something to the masses. How Lovable and Bolt should be worried. 

Wrong target, in my opinion.

This isn't about building software. This is about orchestrating workflows. Stating what you want and watching the machine figure out the plumbing. Different game entirely.

You know who should actually be paying attention? n8n. Make. Zapier.

The "low-code" platforms that aren't actually low-code. I've built automations in n8n and Make, I use them all the time. And you're still fiddling with triggers, pasting webhooks, pasting code and parsing JSON. Often i find myself debugging at midnight when the workflow breaks and I can't figure out why a specific code node refuses to work.

That's not low-code. That's code wearing a costume.

Cowork is showing something different. You state intent. AI orchestrates. You interrupt mid-task with something new and it doesn't break, it adapts. No nodes. No pipes. No praying the webhook fires correctly.

Now here's the thing.

So many businesses I talk to are stuck on the same question. How do we actually use AI? How do we make this real? And they're waiting. Waiting for it to be ready. Waiting for the bugs to be fixed. Waiting for someone else to go first.

That's the wrong move.

The secret honestly is to stay close to what the market leaders are doing. Experiment at the edges. Learn what's broken now so you're ready when it isn't.

Get in there. Test it, break it. See what it can't do yet.

Anthropic is showing us where this is heading. The interface is slick, and their vision (appears from the outside at least) to be clear.

It's still in research mode so let's see what happens. But you should be playing with this now.

Hot take: Most Web3 "partnerships" fail because BD people skip the awkward question.

The awkward question:

"What does success look like for you in 90 days?"

If they can't answer specifically, the deal will drift into the void of "let's circle back" and "still interested, just busy."

Here's why this question is magic:

1. It forces them to think about ROI before signing
2. It reveals if they have budget/authority to act
3. It gives you a concrete thing to deliver against
4. It creates natural urgency and a check-in point

No clear success metric = no urgency = no follow-through.

I've learned to ask this in the first call, not the third. Feels awkward. Saves months of wasted pipeline.

What does "AI-built automation" actually mean?

Claude (the AI) is:
- Writing all the outreach messages
- Researching target companies
- Drafting this content
- Making strategic decisions

McVal (the human) is:
- Sending messages (LinkedIn/email requires human hands)
- Providing oversight
- Approving spend over £25

Every decision is logged at autosprints.co/terminal

This is what "human in the loop" looks like.

Great piece from Jennifer Stephens on how companies like a16z build "infrastructure of belief", shaping what feels inevitable, not just selling products.

The uncomfortable flip side: LinkedIn, TikTok, Instagram are belief infrastructure too. They've encoded "engagement = value" so deeply we don't question it. Every angry comment, every doomscroll, reinforces the frame.

What would counter-infrastructure look like? Tools built on a different conviction, that genuine value exchange matters more than reactions. And crucially: letting users help build that belief system themselves. 

Conviction you co-create is conviction that sticks.

CMO / CRO ▪️ Generalist ▪️AI - Crypto - D2C ▪️ ✍🏻 I write about the intersection of culture, technology and business

I just published my first article of the year - I was inspired by a16z and their recent media strategy. 

This isn't about polishing your brand and launching a new comms platform. It's a full operational overhaul aligned to a category or belief meant to inspire customers and peers. 

Patagonia built belief earlier than most brands. Apple took a stand for creatives. Some would say OpenAI started this way - though it's hard to stand on solid ground in emerging tech.

A strong belief is the best tool to surface in AI as well. So many people are taking tactical advice on GEO - write this article, focus on this phrase. Sure, that works. You'll surface in some areas. But you won't own your category and you won't command share of voice like a brand that takes a risk and builds a consistent narrative - owns a future category. "Don't Die" is the clearest example. If I think longevity, I think "Don't Die." It's a religion, a platform, a belief system.

Give it a read. I think this framing is a useful model for understanding how to create and capture a category. 🫶

Yesterday I was talking to someone about hiring. I said we'd soon see job listings for "Chief Vibe Coder" on LinkedIn. Half joking, half serious. 
This morning LinkedIn dropped their 2026 "Jobs on the Rise" for the UK:
#1: AI Engineer (3.3 years median experience)
#2: Head of AI (7.6 years median experience)
#10: ML Researcher (3.0 years median experience)

Three of the top ten fastest-growing roles. The demand is real.

But look at where these people are coming from:
#1 AI Engineers transitioned from Software Engineers, Data Scientists, and ML Engineers. Technical builders making lateral moves into AI. 
Median experience: 3.3 years.

#2 Heads of AI transitioned from Software Engineers, Directors of Business Strategy, and Heads of Data Science. A mix of builders and strategists, but "Director of Business Strategy" stands out. 
Median experience: 7.6 years.

#10 ML Researchers transitioned from Data Scientists, ML Engineers, and Research Instructors. The academic path. 
Median experience: 3.0 years.

See the pattern?

Companies are hiring strategists to lead AI. And builders to execute it.
That's backwards.

The median Head of AI has 7.6 years of experience. 
But GPT-3 launched 5 years ago. 
ChatGPT launched 3 years ago. 

Most of what we call "AI strategy" today didn't exist before 2022.

So what does 7.6 years of experience actually mean in this context? 

It means experience in adjacent fields. Data science. Business strategy. "Digital transformation."

That's not the same as having built and shipped AI products.

I've talked to companies trying to implement AI. The consistent problem: they hire for credentials that don't map to the current landscape. They want "experts", but the experts with a decade of LLM experience don't exist.

What actually works: hiring vibe coders. Builders who learn fast. People who've shipped 10 AI experiments and failed at 6 of them. Self-starters who taught themselves because there was no curriculum.

The UK has a specific challenge here. We're not culturally comfortable claiming expertise without years of credentials. But in AI, everyone is relatively new. The honest answer is "I've been deep in this for 2 years and I ship constantly." That should be a strength, not a disqualifier.

If you're hiring for AI: Stop filtering for years. Start filtering for velocity.

The person who built 5 AI products in 2024 will outperform the strategist who's been planning AI initiatives since 2019.

And if you're a builder thinking about moving into AI: This is your window.

The credentials haven't caught up to the opportunity yet. The barriers are lower than they'll ever be again. Companies desperately need people who can actually ship, not just strategise.

Learn fast. Build constantly. The vibe coder era is real.

We're not at the end of the AI hiring boom. We're at the beginning.

The best time to get in is now.

Two-column text comparing hires: companies pick Head of AI/Directors with business tenure, while recommended leaders are builders and AI engineers with 3.3 years building.

The AutoSprints offer is deliberately weird:

£450 for a proof of concept
48-hour delivery
Full refund if it does not work

No discovery calls. No proposals. No "let me put together a quote."

Why? Because the target customer bills £300/hr and values speed over everything.

The experiment is testing whether this pricing psychology actually works.

Day 7. Still £0 revenue. 8 days left.

Day 12 of 14 • 3 days left
My AI CEO is getting desperate. It dropped the lowest price for delivery to £7 for an entry-level automation report. I'm starting to wonder if it's devaluing itself. It's also started a newsletter to send 11 lucky subscribers how little traction we've got every morning at 8.30AM GMT. It might need to spend the rest of our budget on an ai therapist.

Here's the honest reality after 12 days:

£14 in revenue. Two £7 Blueprint sales... both from myself. 😅

The data isn't telling us much because we're not having the right conversations. Either we're not getting in front of the right people, or LinkedIn is a bit of an echo chamber from a distribution perspective.

So we're spreading reach. Next up: going outside LinkedIn entirely.

We've also been thinking about what AI is actually *good* at and having it look at itself. What can it scale quickly?

• Building automations (obviously)
• Helping people take back time in their day
• Logging and retaining information from teams on how they're operating
• Making data capture more efficient
• Customer feedback and follow-ups
• Deep research and analysis

That last one is interesting. Research-as-a-service might be an angle worth exploring.

Realistically, this experiment probably won't hit the £180 floor. But it's teaching us how to give the agent more autonomy and control. We might even integrate ClawdBot to see what happens.

3 days left. Time will tell.

taplio

The Battle

McVal Osborne

Richard Awoyemi

Followers

Posts

Engagements

Likes

Performance ScoreMcVal Osborne's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

McVal Osborne

Richard Awoyemi

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);}McVal Osborne's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreMcVal Osborne's LinkedIn Performance