Start here

The founder's playbook for building software in the age of AI

If you are a non-technical founder or leader building a software product, the rules just changed under your feet. AI now writes much of the code, which makes you more capable and more exposed at the same time. The good news is that you do not need to become technical to get this right. The fundamentals did not disappear, they moved, and the job is to understand the new shape of them. This is the whole picture in one place. Start at the top, or jump to the decision you are facing.

In short

You do not need to become technical to build software well. The fundamentals did not disappear, they moved. This playbook covers how to decide what to build, who should lead it, how to staff it, how to keep it good, and how to avoid the expensive mistakes, each section linking to a deeper piece. Start with the decision, not the build.

Begin
  1. What changed, and what didn't

    AI writes code now, fast and cheaply. The quieter truth is what did not change: judgement, leadership, and quality still decide whether a project succeeds. AI commoditised typing, not thinking. The most valuable person on a modern team is not the fastest coder but the one who understands the problem and makes the right calls. If anything, judgement matters more now, because AI makes it trivially easy to build the wrong thing very quickly.

  2. Do you even need to build it?

    The most expensive decision is often the first one: deciding to build at all. Build when the technology is core to your business, proprietary, regulated, or genuinely complex. Otherwise buy it, assemble it with no-code, or have an agency deliver it. AI and no-code have lowered the cost of a first version, so the bar for building from scratch is higher than it used to be. You don't need to build a brewery to drink a pint of beer.

  3. Strategy before tools

    Most failed AI efforts start at the wrong end: a tool, picked because everyone is talking about it, in search of a problem. Start instead from where AI could change your economics, and match each problem to the right approach, which is often not a chatbot. And you can do all of this without a CTO in the room, as long as you have a method and one honest source of technical judgement.

  4. Who leads it?

    You need technology leadership, but not necessarily a full-time CTO. The real question is who owns the technical decisions and the quality, with the authority to say no. A fractional or advisory arrangement gives you that judgement without the full-time cost. But watch the incentives: leadership supplied by the agency that also wants to build for you is conflicted from the start. And do not mistake technical brilliance for leadership, they are different skills, and the second is the one that decides outcomes.

  5. The team, and the hiring traps

    However you staff it, the people decisions are where money quietly leaks. Recruiters are paid when people move, not when they stay. Agencies showcase A-players and deliver juniors. The defences are the same everywhere: name people in the contract, do your own diligence, and never outsource the judgement. Treat suppliers as suppliers, with incentives of their own.

  6. The quality you can't see

    Software has a visible quality, does it work and look good, and an invisible one, does it scale, stay secure, and survive being changed. The invisible kind is what silently kills companies, usually at the worst moment: a fundraise, a launch, an exit. AI makes this more urgent, because it is excellent at the quality you can see and weakest at the quality you can't.

  7. Why projects fail, and how to recover

    Most AI projects that fail do so for predictable reasons, and most failures are recoverable. The causes are rarely the technology: a use case chosen for excitement, no success metric, the unglamorous engineering skipped. If you are already in the hole, the first move is an honest diagnosis, ideally from someone who did not build the thing that failed.

  8. Governance and accountability

    As AI moves into real decisions, the question of who is accountable when it goes wrong stops being theoretical. You do not need a Chief AI Officer. You need a named owner and proportionate governance: an inventory of what is in use, clear lines on what AI may decide alone, and a simple record. Regulation is tightening, and 'the board' is not an answer.

  9. What's coming

    None of this is static. AI is shifting from tools you drive to agents that work on their own, the job market is being rebuilt around them, and the durable advantage is moving to the things machines can't do: trust, judgement, taste. Building well today means building for that direction, not just this quarter.

Where to start

Start with the decision, not the build. Work out whether you should build at all, what problem is genuinely worth solving, and who you trust to pressure-test the plan. Everything else follows from getting those three right. If you want an independent read on any of it, with no stack to sell you and nothing to upsell, that is exactly what I do.

Get in touch

Frequently asked questions

Do you need to be technical to build software?
No. Building a software product is a business exercise as much as a technical one. You need a method, plain business judgement, and a little independent technical input at the few points that matter, not a computer science degree.
What has AI changed about building software?
AI now writes much of the code, which makes a non-technical founder more capable and more exposed at once. It lowers the cost of the easy parts, not the hard ones (scale, security, quality), and it does not remove the need for judgement and leadership.
Do I need a CTO to build a tech product?
Not always a full-time one. You need impartial technical judgement at the key decisions, which a fractional or advisory arrangement often provides more cheaply, and without the conflict of interest of asking the vendor who will build it.
How do I avoid wasting money on software?
Start from value not tools, validate cheaply before building, choose build-vs-buy on whether the technology is core, get one independent read before committing, and treat agencies and recruiters as suppliers with incentives of their own.
Where should I start?
With the decision, not the build. Work out whether you should build at all, what problem is genuinely worth solving, and who you trust to pressure-test the plan. The sections of this playbook walk through each step in order.
Stay ahead

AI & tech are moving fast.
Get the signal, not the noise