AI is a guidebook. You need a local.

We've all been there. A new place, exciting, but no idea where to start. Which beach is the best beach? Are there any 'secret local spots'? Where do the locals eat (but really, where...)?

A search or an AI chat will answer in seconds, and the answer will look right. But, part of you still wants to ask a local, because only a local's word actually puts your mind at ease. That instinct? It's spot on.

Here's my point. Founders are tourists in most of what we do. You're a marketer, or a designer, a researcher, an engineer, a go-to-market person. And to take it further, you're a marketer with experience in one, maybe two verticals. One spoke on the chart runs long. The rest? Stubs.

Nobody arrives an expert in all five. That's what a founder is. One deep skill you've spent years building, and four more you're expected to cover anyway. Your skill is that one town you know street by street — but on the other side of the world, there's a lot of unfamiliar ground.

The question is, what do we do when we're the tourist in someone else's backyard?

The instinct today is to ask AI, and in your home town (or even your state) it's excellent.

Point AI at a question in your line of expertise and it's genuinely very good. You read the output and within seconds you know it's right, or you know exactly which part isn't. You correct it, you move on, and you're several times faster (and more confident) than you were before.

And that experience is real. It's why 80% of the 1,719 people McKinsey surveyed across 97 countries said AI had improved their productivity.

Now, ask it about a place you've never lived.

You get an answer that reads exactly the same. Confident, structured, well organised. The difference? You've got no way of telling whether it's right, because you've never lived there.

The local spot and the tourist trap read exactly the same.

They stay that way until you commit to the 8km walk, only to find every other tourist has done exactly the same thing. Everyone used AI to 'outsmart' the crowd, and it led them all to the same beach — while the locals quietly moved to the other 'secret local spots'.

AI can supply judgment. It can't supply the confidence, experience (failures) and expertise that come with human judgment.

That's the whole thing, really.

The output is much the same quality either way. What changes is our ability to evaluate it. In the long spoke, checking is instant and free. In a stub? Not available at any speed.

Which means we get faster on home turf, and more confidently wrong everywhere else.

And that's where founders are getting stuck. We're pushing our judgment to its limit every day, because with AI the expectation is now that we're good at everything — covering every base and more. The decision fatigue is real, and it's taking its toll.

Isn't this just about asking the right question?

I'm not saying the AI is completely incapable. What I am saying is:

1. Better prompting doesn't always fix this. Knowing what to ask is local knowledge too — a prompt before the prompt is still your judgment, on ground you've never walked.
2. Agents often make it worse. They run on your assumptions and check in less often, so every hand-back leans harder on judgment you don't have.

There's no prompt for experience. But take a local who knows how to harness AI in their field of expertise, and you've got the guide you need.

McKinsey's latest survey shows what all of this adds up to at scale. 80% felt more productive. 37% said AI had contributed anything at all to their company's EBIT, unchanged from a year earlier, while enterprise-wide adoption rose from 38% to 44%.

80%

Report a personal productivity gain from AI. McKinsey, The state of AI in 2026, 25 August 2026, n=1,719.

37%

Report any contribution to their company's EBIT at all — unchanged on a year earlier, while enterprise-wide adoption rose from 38% to 44%.

Individual speed went up. Business outcomes didn't move. That's what a hidden gap looks like in a spreadsheet.

And we can't close it ourselves, at least not quickly.

Sure, we can read about pricing. What we can't do is acquire the experience of having priced something wrong and watched the churn come through.

Human judgment is compressed consequence. Years of being wrong and paying for it. That's the only way it gets made. It can't be scaled with compute, it can't be read into existence, and it can't be bought forward.

Which is why the blind spots stay blind for as long as they do. It's why things always look different in hindsight — we were warned, we were told — we chose to do it anyway...

The other option: ask a local.

We asked 84 senior professionals about contributing their expertise to startups. The locals aren't just out there — they're keen.

68%

Highly interested (4–5 out of 5) in earning startup equity for their expertise while keeping their current salary.

92%

Would commit regular weekly time — a few hours to a few days — for the right amount of equity.

73%

Already in full-time work: exactly the people a job ad will never reach. GetSweaty Professionals Survey, September 2026, n=84.

Fill the gap with a local — someone whose home town is the place you're guessing about. Then hand them the guidebook.

Three things make that practical now in a way it just wasn't a few years ago.

1. You need their judgment, not their week. They're exited founders, eager to pass on their knowledge, their 'judgement'. Scope the engagement to the decision and it costs a fraction of what a hire would. What is more appealing to them is ownership, a small piece of something they help you to build.
2. The people with those long spokes aren't looking for jobs. They're employed, senior, well paid, and fifteen years deep in a domain you've never worked. You're not hiring a head of sales full-time. You need someone who has priced this kind of product before to spend 4 hours a week for 3 weeks on the specific pricing decision sitting in front of you. They aren't on the market, and a job ad won't reach them.
3. Ownership is the currency that matches. It's the only thing a founder holds that's priced in the same units as what we're short of, which is future consequence. That's the whole design of GetSweaty. Micro-equity, 0.05% to 5%, vesting against milestones rather than time. The legal, the tax compliance? Taken care of, in Australia and the UK.

The part that compounds.

Here's what actually changes once a local covers the ground you were guessing on.

You don't just get one good decision out of them. You get the ability to use AI properly in that domain, because there's finally somebody who can tell the local spot from the tourist trap.

The multiplier that only worked on your turf starts working on theirs.

A guidebook in the right hands is a different tool. That's the version worth aiming at: the person best placed to use it, in the places where you can't — so it gets you where you need to be, with the confidence and peace of mind to match. Whilst also allowing you to refocus back on the unique skills that got you here in the first place.

Where this could be wrong.

There's a fair objection to all this, and Anish Acharya put it sharpest on the a16z podcast in August: locals can be out of date too. Senior founders who haven't survived a recent product cycle, he said, are held back by distance from the technology. They carry a sense of the ceiling that's rooted in the past.

And look, he's right — some locals have been away from town for a while. The tech streets get rebuilt every product cycle, and a five-year-old map will walk you into walls.

But notice which map that is. The map of what customers value — and what they'll pay for — barely changes. That's the ground a local still knows, however long they've been away.

So ask your locals about the ground that doesn't move so swiftly; customers, pricing, what people pay for — not the vision. The vision? That's your turf. Keep it.

The guidebook isn't going to make you a local. And asking it harder mostly makes the traps better disguised.

But somebody out there has spent fifteen years in the place you've just landed. You can't afford their time — you can afford to share what you're building. A small slice of equity, scoped to the decision in front of you.

It's not that AI isn't capable. It's that you don't need to be everything in the business, all the time. You've got levers to pull — experts whose judgment makes the business significantly better, and makes you faster and more confident alongside them.

So here's the move: pick the domain where you're guessing the most, and go find your local on GetSweaty.

Besides — the adventure shared is often more fulfilling than the one taken alone.

Sources

1. McKinsey & Company, "The state of AI in 2026: On the road to ROI", 25 August 2026. Fielded 4 May to 8 June 2026, n=1,719 across 97 nations. mckinsey.com
2. Anish Acharya on the a16z podcast, late August 2026. Verified against the episode transcript, 8 September 2026.
3. GetSweaty Professionals Survey, fielded September 2026, n=84 (self-reported; respondents across Australia, the UK, North America and beyond).