What do venture capitalists look for now that AI makes building cheap?

Ask a room of founders what venture capitalists look for and you'll hear the familiar list: traction, a big market, a strong team. All still true. But when the cost of shipping a first version is heading towards zero, that list gets reshuffled. If anyone can build the thing, what's actually left to back?

We put that question to Oliver Darwin, an investor at OIF Ventures, on Now or Never. OIF writes $2 to $5 million cheques out of a $140 million fund and runs a high conviction, concentrated portfolio, so it can only take so many bets. Oliver's answer was refreshingly direct. The founder matters more than ever, and the two things he keeps coming back to are domain experience and taste.

What do venture capitalists look for when anyone can build?

Oliver describes OIF as founder led. The founder has always carried the most weight in how the fund assesses an investment. What's changed is how much now rides on that. With AI tools, engineers (and everyone else) are far more efficient, a seed round stretches further than it used to, and getting to a V0 costs a fraction of what it did. Oliver even raises the idea of "seed strapping": raising one round that gets you all the way. When the build is cheap, more of the value sits with the person deciding what to build.

"Creating the thing is the easy part now. It's how you create the thing or even should you create the thing."

That's a useful lens if you're preparing to raise. Your demo proves you can ship. It doesn't prove you should have shipped that, or that you made the right calls about how.

Domain experience: knowing the workflow better than anyone

Oliver's view of software is old school in a good way. Software has always been an opinionated, best-in-class version of a workflow, delivered through UI and UX, and he doesn't think AI changes that. His worry is that many AI products have forgotten why people buy software in the first place: to remove the ambiguity of how to do something.

Domain experience is what lets a founder map that workflow properly. It tells you where the product has to be deterministic and where generative AI can amplify what's already there. Compliance is his example: you can't have a model improvising an output where getting it wrong matters. He points to Relume as a team that gets the balance right. The founders are designers, they know the workflow intimately, and a large share of the product runs on deterministic logic with AI amplifying what they're already good at.

He tests for this in conversation. How are you thinking about the problem? Why would or wouldn't the incumbent build this? He wants to see that the founder has thought deeply, rather than outsourcing the thinking or taking the convenient option. Being honest about what you don't know helps rather than hurts:

"To be able to understand what you don't know actually is an incredibly good indicator of your depth of understanding of your domain."

Taste: hard to measure, impossible to fake for long

Taste is hard to pin down, and Oliver admits as much. But he's specific about how it shows up in a product: as discretion.

"If you're a founder that has taste, you will naturally also exercise discretion as to what you build and how you build it."

The opposite is what he sees a lot of right now: solutions looking for problems. Because building is so easy, people build something rather than being discerning about what they've built. Does this need to be AI? Often, probably not. One OIF portfolio company helps councils run rates payments and deliberately doesn't use AI in its product. Its edge is trust with government and ratepayers, and introducing uncertainty would erode exactly that.

He applies the same thinking to his own job. He uses AI as a thought partner and a critic of his own opinions, but he's wary of handing over the thinking itself. Conviction, in his view, isn't something you can outsource.

Why design is now part of what investors look for in a startup

This is the part B2B founders should sit with. Oliver sees a consumerisation of B2B. Products now go viral on LinkedIn and X in a way Salesforce never did, and OIF sees businesses that have reached $2 to $5 million in ARR so quickly they haven't been through a real churn cycle. That makes revenue harder to read. It can look more like consumer revenue: higher churn, lower retention, harder upsell.

When the numbers are noisier, the experience carries more weight. In consumer products, the time from the start of onboarding to the first point of value is make or break, and he's watching that pattern move into B2B. He named Superhuman and Granola as products where the design does the selling. The alternative he described will be familiar: a sales rep booking you in for a 30 minute call just to check you know how to use the tool.

"If you can design your way around needing a power user in an organization, that's the holy grail."

First impressions also stick. If an ideal customer tries your product and has a poor experience, they're unlikely to come back in six months just because the roadmap improved. Oliver thinks plenty of companies are leaking ICP customers this way.

Design is distribution, and storytelling

The blind spot he sees most often: founders underestimate how tightly design links to distribution and to the way they tell their story. He's tried to introduce founders to potential customers and found they didn't have so much as a paragraph on what they do, plus a two year old website that doesn't say it either. His lesson from investment banking applies:

"If it looks good, inherently people will trust it and think it's good."

Vouch, another OIF portfolio company, is his example of distribution built into the product. Anyone watching one of its video testimonials can see it was made with Vouch, the same way you can't miss an email sent from Superhuman.

This matches what we see at UntilNow. AI has made execution cheap, which is exactly why UX, taste and brand matter more, not less. When ten teams can ship a similar product in a month, the one that's clearer, quicker to value and easier to trust tends to win. That's the job of digital product design, and investors are increasingly judging it.

Customer obsession is an investor signal

Most people who churn won't tell you why. Oliver admits he rarely does. So someone has to go looking, and that's why he rates customer obsession so highly. He wants founders who actively chase down why people leave and what they could fix.

He also reminds founders how close to the coal face the tech world is. Most users don't care what technology sits underneath. They want a piece of friction removed from their day. And because OIF always talks to customers during due diligence, what your customers think of the product ends up in the investment decision either way.

What Oliver changed his mind about

Two things, and they pull in slightly different directions. First, the best product doesn't always win. You need a good product, but distribution is usually what makes the difference. Second, he's stopped worrying about whether Salesforce or Shopify will build the same thing. Coming from banking, where mature incumbents dominate, that took some adjusting.

He's now the least receptive person on his team to "it's a competitive market" objections, because he's come to believe that "execution and taste are the only things that matter."

What to take into your next raise

• Know the workflow better than anyone in the room, and be upfront about what you don't know yet.
• Show judgement about what you chose not to build, including where you didn't use AI.
• Treat onboarding and time to first value as part of your metrics story, not a design detail. If you're early, our guide to UX design for startups covers what to do first.
• Get your story straight: one clear paragraph, and a website that says what you do.
• Be visibly customer obsessed. Investors will talk to your customers anyway.

Oliver's point is simple: when anyone can build, judgement is the asset. For the full conversation, including his take on services roll-ups and how OIF uses AI in its own process, listen to the episode with Oliver Darwin on Now or Never.