In the AI era, the crucial questions all founders dread are more important than ever

Virginia Murdoch was building software well before it became easy.

In 2011 she co-founded Booki.sh, a web-native ebook reader that was well ahead of its time and anticipated reader concerns about Amazon and Apple’s ecosystem lock-in.

It was acquired by OverDrive, then the leading global distributor of ebooks and audiobooks, after about three months in the market and with about 150 users. So when she told a room of founders at Google Ascend in Sydney last week what it’s like to build with AI now, people listened.

Vibe coding, she said, makes you feel like a genius. You describe what you want, the code flows, and within days you’ve got something that works. Then you put it in front of a customer and they say, “I don’t exactly know what you’re doing here.”

The cost of building software has dropped to almost nothing. Ten years ago, the founder who could build the best thing fastest usually had the edge. Now anyone can vibe-code a product over a weekend, so that advantage is mostly gone.

The customer test

What hasn’t got any easier is finding out whether anyone wants it. That still means sitting across from customers, asking the right questions, and being trusted enough that they answer honestly. Many founders I meet would rather do almost anything else.

If anything, AI has made it harder. When a build took months, there was time for doubt to creep in. Now the product shows up so fast, and seems to do so much, that you fall for it before anyone else has seen it. And the more attached you get, the less you want to hear that the problem it solves isn’t real.

Years ago, when Virginia helped my private equity team get a global website to cater to the over 50 countries where it was sold, the build took months. Today it’s an afternoon’s work. The barrier to building fell so now founder discipline must be laser focused on the customer.

Questions founders avoid

“Talk to your customers” might be the most repeated advice in startups. You hear it at every pitch night. What nobody tells you is what to actually ask.

The obvious question, “what do you want?”, is close to useless. Customer’s mostly don’t know. They know what annoys them and what they’ve put up with for years, but they can’t design your product for you.

Then there’s the other problem. Fear. A lot of founders don’t want to ask let alone hear the answers.

So we asked the room three questions:

• What does your customer do today instead of using your product? The honest answer is often nothing, or a manual workaround they’re perfectly happy with.
  
• What would they have to give up or change to switch? Another app or another subscription costs the customer time, attention and effort, which are scarce. Founders can forget that.
  
• What’s your riskiest assumption, and what’s the cheapest way to test it? Most founders already know which assumption they’re avoiding.

Trust comes first

Virginia made one more point I keep coming back to. The question used to be who could build the best thing fastest. Now it’s whether you are the person customers trust to solve their problem.

There’s so much noise, and so many people chasing the same credit card, that a good product isn’t enough to cut through. Canva had room to grow when the default was Adobe.

Today, when someone has a problem, they often think first of the person they already trust on that topic, sometimes before any product exists.

That trust comes from showing up and being useful long before you ask for anything.

• Will Richardson is the managing partner of Giant Leap, Australia’s first impact venture fund.