3X win rate on "we'll build it ourselves" deals.

AI researcher turned founder turned growth engineer. Still figuring it out, in public.

M.Sc. in deep learning. Former AI Team Lead. Published. Then I got hungry for more impact and founded The Voice Box, my AI agency.

Needed leads, so I went deep on LinkedIn: 40,000 followers in 8 months, $100K closed in 90 days. All organic.

Growth is a blend of experimentation, data, and human psychology, increasingly engineering.

Now I'm the founding growth engineer at Swan, building the first autonomous business where humans are amplified 100x.

This is where I unpack it as it happens.

Stick around.

A $1B CRO told me he could build a Swan equivalent with Claude.

I agreed...

We 3X'd our win rate in the last 30 days on deals where the prospect said "we can build this with Claude."

Not by arguing... By agreeing with them.

Here's the exact reframe so you can run it too.

Every vendor I talk to is hearing the same sentence in deals right now:

"We can probably build this ourselves on Claude."

And most sellers panic, because for the first time in twenty years the objection is technically true.

The code is no longer the moat. A smart builder with a Claude seat can actually ship a version of your product.

So stop defending the code... You were never selling it.

Nobody in the history of enterprise software ever bought development hours.

What a buyer actually pays for is the thousands of decisions someone else already made, and already got wrong, before they showed up.

Which architecture survives contact with messy CRM data.

Which edge case quietly poisons everything downstream.

What to do when the model is confidently wrong..

That's the product. The code is just the residue of those decisions.

And here's the part your buyer hasn't priced in:

Enterprises are structurally terrible at making decisions.

It's not a talent problem, it's a design problem. Committees, sign-offs, competing owners, quarterly planning cycles. An enterprise is built to avoid one bad decision, not to make a thousand fast ones.

A startup's real unfair advantage was never talent or technology. It was decision velocity.

So when they say "we'll build it on Claude", they're not comparing your price to their dev cost.

What they're actually buying at Swan AI, and at yours:

→ De-risk: every decision in the product already failed somewhere else first, on someone else's budget.

→ Speed: they skip the thousand-decision queue and start at the finish line.

→ Outcome: they get the version that survived hundreds of customers, not their team's first guess.

Faster time to value, at a fraction of the risk, with a better bottom line.

Who can say no to that?

So next time a buyer says "we can build this on Claude", try this:

"You're right, you can. The question is whether you want your team making the first thousand decisions, or inheriting the ones that already survived."

-Ariel

The Hypothesis

Buyers never paid for the code

A builder with a Claude seat can ship a version of your product. What they can't skip: the thousands of decisions someone else already got wrong.

Your Data Point

What's one mistake your product already made so your customers never have to?

Next time someone says "we'll build it on Claude," just agree. Then reply and tell me what happened :)

Community Notes

Skip a few hundred decisions: 334 free skills from operators who already made them here.

getswan.com