How to Find a Technical Cofounder in the AI Era

The number one question I get from entrepreneurs, by a distance, is how to find a technical cofounder.

Not how to raise money. Not how to price. How to find the person who will actually build the thing. I get asked roughly three times a day, ten times a week, just in my own small corner of the world. It is the most common question in software entrepreneurship and it is the one with the least satisfying answer.

Here is the honest version, and it starts with a number most founders have never been told.

Why half of the best engineers will never join a startup

Before you make a single call, understand the size of the pool you are fishing in.

Roughly half of the engineers with the skill to be a real technical cofounder will not join a startup under any circumstances. Not yours, not anyone’s. Their spouse, their own temperament, or their life experience has led them to the conclusion that the risk is not worth it. They want a company that pays a strong salary and good benefits, and that is a perfectly rational choice.

The opportunity cost is brutal. One of the best engineers I know claims he earns three quarters of a million dollars a year. His view of startups is simple: every year or two at his current job is the financial equivalent of building a startup and selling it, without the probability of failure. Why would he ever do a startup?

Large technology companies need that talent as badly as you do, and they have the capital to pay for it. So you are competing for the half of the population that is open to the idea at all, against employers with far deeper pockets.

In good economic times, this is probably the single biggest thing holding back software startups. Not ideas. Not capital. Finding one person who can get along with the business founder, write most of the version 1.0 themselves, and ship it to the first paying customers.

That is a tall order. And it is worth saying plainly that you want one excellent engineer at the start, not a crowd. There is a reason The Mythical Man-Month has stayed in print since 1975: adding more engineers to an early project tends to make it slower, not faster. You do not want a committee of mediocre developers. You want one person who genuinely knows what they are doing.

Do you still need a technical cofounder in the AI era?

Everything above has been true for twenty years. Some of it stopped being true in the last two.

It is worth being precise about which parts, because founders are currently drawing the wrong conclusion in both directions. Some have decided they never need a technical cofounder again. Others are recruiting exactly as they would have in 2019. Both are wrong.

What genuinely changed.

You can now build the first version yourself. Not a mockup, not a clickable prototype, a real product that real customers pay for. AI coding agents have moved a competent, determined non-technical founder from “cannot build” to “can build slowly and imperfectly,” and imperfectly is enough to find out whether anyone wants it.

Learning enough to be dangerous is also easier than it has ever been. For a meaningful number of founders, becoming technical enough to own the early product is now a live option rather than a fantasy. That was not true when I started giving this advice.

And the thing I said earlier about traction being your best recruiting tool now costs a fraction of what it did. The founder who used to arrive with a wireframe can now arrive with working software and paying customers. Your negotiating position with a strong engineer is dramatically better than it was, and you can reach it alone.

What did not change at all.

Half of the best engineers still will not join a startup. If anything AI raised their market value and made their day job more pleasant, so the opportunity cost argument got worse for you, not better.

Compensation reality did not change. The excellent engineer who will work for nothing still does not exist.

The moonlighter still fails on the same schedule.

And you still cannot evaluate technical skill yourself. This one actually got harder. A candidate with a good agent produces impressive-looking output on a take-home exercise regardless of whether they understand it. The twenty-minute whiteboard conversation with someone who genuinely knows the craft is more necessary now than it was, not less.

Most importantly, the gap between you and a professional engineer did not close. It probably widened. The same tools that moved you from zero to something also moved a strong engineer from fast to extraordinarily fast, and a multiplier applied to a much higher base produces a much bigger number. When an experienced engineer works with these tools, they know which output to trust and which to throw away, they recognise a bad architectural suggestion before it becomes six weeks of rework, and they write the hard parts by hand when the model is clearly guessing. You cannot do any of that, and no amount of prompting substitutes for it. You are not competing against the version of your competitor who had no engineer. You are competing against their engineer, amplified.

There are three specific things you cannot assess in your own AI-built product, and they are the three that matter: whether the code quality will hold as it grows, whether it is secure, and whether it has any taste. Security is the one that should worry you most, because it fails silently and you have no way to know. A product can take payments for a year before anyone discovers what was wrong with it.

The new failure mode nobody warned you about.

Here is the thing I now watch founders walk into.

AI makes it easy to accumulate a large amount of code that nobody has ever understood. Not badly written code necessarily, just code that no human has held in their head, built in a direction no one chose deliberately, with decisions nobody can explain.

I used to tell founders that the next engineer never wants to use the previous engineer’s code base. That problem did not go away. It got bigger, because the volume is bigger and the authorship is thinner. When you finally recruit your technical cofounder, what you hand them can be a liability rather than an asset, and a strong engineer will see that in the first hour.

The second version of this is worse. AI lets you build the wrong thing much faster. If you have not validated the business, all the speed does is get you to the dead end sooner, with more code to abandon when you arrive. Speed without clarity creates chaos, which is a thing we have written about before in the context of how we invest.

There is a subtler cost here that I think is underappreciated. When a hastily built product fails, founders almost always conclude the idea was wrong. Sometimes it was. But sometimes the product was slow, broken, or ugly enough that customers left for reasons that had nothing to do with demand, and the founder has now killed a good idea on bad evidence. A product you cannot evaluate produces a signal you cannot trust, which defeats the entire purpose of building it in order to learn something. We have never had a shortage of products nobody wanted. What is new is the growing pile of perfectly good ideas that were buried by their own first version.

So what should you actually do differently?

The sequencing has genuinely inverted. The old order was recruit a technical cofounder, then build, then find customers. For most software businesses in 2026 the order is:

1. Validate the business model with no code at all
2. Build the first version yourself with AI, deliberately and in a technology a real engineer could live with
3. Get paying customers
4. Recruit your technical cofounder from a position of strength, with revenue, a product, and far better terms than you could have offered before

The question “how do I find a technical cofounder” is often being asked at step one when it belongs at step four. That is the single most useful update to everything I have said about this over the years.

What has not inverted is the ending. If the business works, you still need someone who can own the architecture, and you need them sooner than feels comfortable. AI raised the ceiling on what you can do alone. It did not remove the ceiling.

Validate the business before you recruit a cofounder

Before any of the recruiting advice, there is a question you should ask yourself that most founders skip entirely.

Should you be looking yet?

Before I am looking for a technical cofounder, before I hire any kind of team, before anyone writes a single line of code, I want to have gone through real validation. Product-market fit. A business model where I can create, capture, and deliver value. Proof that this is a business and not a hypothesis I have grown attached to.

Writing code against an unvalidated business model is a form of waste, and it is the most expensive kind. Spending six months building something that any competent advisor could have told you was dead on arrival destroys the only resource a founder can never get back, which is time.

This is the entire reason our 3-day bootcamp exists, and why it runs in three days rather than three months. Founders are talking to real customers by day two. A long build cycle gives a team room to look productive while learning nothing, and the compression is deliberate.

There is a second reason to validate first, and it is purely tactical. Traction is the most persuasive thing you can put in front of a strong engineer. It works on them the same way it works on investors. Walk in with wireframes, a validated business model, letters of intent, or actual signed contracts, and you are no longer a business person asking someone else to do the hard part. You are someone who has already done their half of the work.

I have watched founders take real checks for a product that did not exist yet. The reasoning holds up: the worst case is that you give the money back. That is the kind of evidence that changes a recruiting conversation.

Startup Ignition Bootcamp 3 days. In person. 0% equity. 1,000+ ventures have come through and gone on to build real businesses. View Cohorts & Apply →

Finding a technical cofounder is a numbers game

Let me tell you about a company that lost its technical cofounder.

They needed a new one. I had lined up an excellent angel investor who was willing not only to invest but to come in and help lead alongside the founders. The one condition was that they find a technical cofounder. Weeks went by. Then months.

Eventually that investor and I sat the two of them down and asked what was going on. They told us they had tried everything.

I told them I did not believe them. I said that if they did not find somebody within seventy-two hours, the investor was walking away, his money was leaving, the offer was gone, and I would never work with them again.

They were upset. They said that was a terrible thing to say.

They found someone in about sixty hours, with twelve to spare.

I am not telling that story to recommend threatening people. I am telling it because of what it reveals. They had not tried everything. They had tried the comfortable things. Properly motivated, with the only alternative being complete failure, they did in two and a half days what they had failed to do in months.

I have seen the identical pattern with salespeople. A rep tells me there is no way to double sales next month. Then circumstances change and they do it. Humans are remarkable when the stakes are real.

Business is a numbers game. If you want ten sales next month you do not make ten sales calls, you make a hundred. Finding a technical cofounder works exactly the same way, and most founders who tell me they have exhausted their options have made about ten calls.

Where to find a technical cofounder

Turn over every rock. Be a pied piper for your own vision. Become a genuinely good networker, not a person who attends one event a quarter.

In the Utah market where I spend most of my time, there is some kind of meetup, event, or gathering where business and technical people mix nearly every night of the week. Most startup markets have some version of this. Go to them constantly.

The highest-leverage move is the one almost nobody makes. When you find an engineer with exactly the skills you need and they turn you down, do not end the conversation there. Ask them this:

Who do you know who is like you? Who among your friends would love a founding stake in a great venture?

Strong engineers know other strong engineers. A no from the right person should produce three new names. If it does not, you ended the conversation too early.

How to vet a technical cofounder when you cannot code

This is the part where non-technical founders are most exposed, and it took me a long time to solve.

Engineers vary enormously in ability, and some of them, good and bad, will overstate their skills. You will interview someone who tells you they are an expert in Ruby on Rails, or Python, or React, and that evening they are reading the beginner’s guide to the thing they claimed to have mastered. You have no way to tell the difference. Confidence reads as competence when you cannot evaluate the work.

My answer has always been to keep people in my life who can call it. Computer science PhDs who will spend twenty minutes with a candidate I send them, put them at a whiteboard, make them write real procedures, and then tell me thumbs up or thumbs down. It is the only reliable signal I have ever had.

Sometimes the verdict is blunt. I have been told plainly that a candidate was not as good as they presented. Once I was told the opposite: that the person in front of us was the best engineer he had seen in the state. We gave that person eight percent of the company, as roughly the third or fourth engineer in. He was worth it.

If you do not have someone like that in your network, there are paid technical interview services that will run a candidate through a panel of legitimate engineers. Compared to the cost of a bad technical cofounder, it is inexpensive.

What you must not do is trust your own read.

Should a technical cofounder keep their day job?

If there is one pattern I would most like founders to stop repeating, it is this one.

A founder calls me, delighted. They have finally found their technical cofounder. The person will not quit their job, but they are going to put in two to four hours every evening and get the MVP built.

I now know roughly how this ends and roughly when.

About a month in, give or take, once the honeymoon is over, that person comes back to the founder who is all-in and full-time on this company and explains that their workload at their day job has increased and they cannot continue. The founder starts over.

I have watched founders cycle through three, four, five people like this inside six months. And it compounds, because the next engineer almost never wants to build on the last one’s code. The volume of software written and thrown away to false starts in this industry is astronomical.

A person doing two to four hours a night after a full day of work is not your cofounder. They are a temporary contractor with no contract, and you should plan accordingly.

How much equity should a technical cofounder get?

Here is the fantasy I hear most: a genuine 10x engineer who will build version 1.0 essentially alone, for no salary, in exchange for around eight percent of the company.

That does not happen. It is a pipe dream.

And in the rare case where a founder talks a naive strong engineer into those terms, it blows up in both their faces within a month or two, once the engineer talks to a friend or runs the numbers.

It has to be one of two things: a healthy cofounder-sized chunk of equity, or real salary compensation. The market rate for a genuinely excellent engineer, especially somewhere like Silicon Slopes, reflects the fact that they can earn a great living anywhere and have their pick of employers. Joining your startup is not a small choice for them.

What the number should be depends almost entirely on when they arrive, and this is where the sequencing I described earlier turns into real money. The rough ladder most of the market runs on: someone who starts alongside you on day one, sharing the full risk, is near an even split. Someone joining after you have built and shipped an MVP is meaningfully below that. Someone joining a company with real revenue and customers is lower again. Four-year vesting with a one-year cliff is standard at every point on that ladder, and you should insist on it for yourself as well as for them.

That spread is the argument for validating and building before you recruit, stated in the only terms that are unambiguous. The founder who shows up with paying customers is not just more attractive to an engineer. They are retaining materially more of their own company.

Understand the levers of salary and equity, and do not live in dreamland about either.

One related warning. Founders often ask about recruiting an engineer plus their best friend, or a pair of developers instead of a single technical cofounder. I have seen people recruit a college roommate or a best friend into C-level roles repeatedly, and almost invariably it has blown up. If you do bring in two technical people of roughly equal standing, the question is whether you as the business founder have the competence and the confidence to manage them. If you cannot manage a technical situation at all, two of them will not fix it. If you genuinely can lead technical people, it can work.

Alternatives to a technical cofounder, and their limits

If you cannot find the right person yet, you are not stuck. There are ways to make progress, and all of them come with the same caveat.

You can hire a contract team to build a wireframe or an MVP. You can work with a student team. You can use offshore development. You can use no-code platforms, and the modern ones are genuinely capable.

You can also rent the judgement rather than the hands. A fractional or part-time CTO costs real money monthly rather than equity, and the point of one is not that they write your code. It is that someone competent is reviewing what you and your AI tools produced, catching the security and architecture problems you cannot see, and telling you the truth about what you have. It is also far easier to end that arrangement than to remove a cofounder from your cap table, which is worth something on its own. With nothing but yourself and a designer you can build a functioning prototype of a fairly mainstream product, and that is sometimes enough to reach five or ten thousand dollars a month in revenue, which is real evidence.

Some people call this blueprinting, and the analogy is a good one. You would not put up a building without blueprints. Do not build a product without them either. Come to a contract team with the business model already figured out from your validation work, have them wireframe the product, and now you have something concrete to put in front of a real technical cofounder.

Two cautions.

If you go this route, do the blueprinting properly. Founders come out of our bootcamp with a tested business model and a clear picture of what the first version actually needs to do, which is exactly the input a contract team needs and almost never gets.

First, insist that whatever gets built uses technology your eventual in-house team can live with. Code that has to be thrown away when the real engineer arrives was not a bridge, it was a detour.

Second, and more important, these are all ways of punting the question rather than answering it. In my experience, if the business model is genuinely going to go somewhere, you reach the point of needing an in-house technical cofounder in a hurry. The traction that proves you were right is the same traction that makes the gap urgent.

If you are validating before you build, the AI ToolSuite is designed for exactly this stage, so a non-technical founder can test the business model without waiting on an engineer.

How long it takes to find a technical cofounder

This is the piece of advice I give most often and that founders most often ignore.

There are two parallel tracks, and you must be on both at once.

On the first track, you are validating the business model, designing the MVP, and possibly working with a contract team to get to a wireframe or version 1.0.

On the second track, starting the same day, you are looking for your in-house technical cofounder. Continuously. Whether or not you need one this month.

What happens otherwise is predictable. Founders find the search so unpleasant that they simply do not do it. They go down the first track alone, traction starts to arrive, things get exciting, and only then do they discover that landing a strong engineer in-house takes three to six months. They are now half a year behind at the exact moment the business is taking off.

Always be finding. And always assume your technical cofounder will not be there tomorrow.

The short version

If you take nothing else from this:

Half your candidate pool is unavailable at any price, so the search is harder than you think and you are probably not trying as hard as you believe you are.

Validate the business before you recruit anyone, because traction is both the thing that de-risks your company and the thing that makes a strong engineer say yes.

Never accept a moonlighter as a cofounder. It fails on a schedule.

Get someone technical to vet technical skill, because you cannot.

Be honest about compensation. The free 10x engineer does not exist.

Use AI to build the first version yourself, then recruit from a position of strength. The sequencing inverted, and the founder who arrives with customers keeps more of their company.

And run both tracks at once, from day one, because by the time you obviously need this person you are already six months late.

If you want help with the first step, that is what we built the Startup Ignition Bootcamp to do. Three days, in person, no equity, and you leave knowing whether the thing is real before you go asking anyone to build it.

Frequently Asked Questions

How do I find a technical cofounder?

Treat it as a numbers game that you run aggressively, not a search you do passively. Work your existing network first, then ask every strong engineer who turns you down for introductions to people like them, and show up at the places technical people already gather. Expect it to take three to six months, and expect to talk to far more people than feels reasonable. Most founders who say they have tried everything have made about ten attempts when the job needs a hundred.

How much equity should a technical cofounder get?

Enough that the arrangement survives contact with reality. The common fantasy is a genuinely excellent engineer who works for no salary in exchange for around eight percent. That does not happen, and on the rare occasion someone talks a naive engineer into it, it falls apart within a month or two once the engineer does the math. It has to be either a healthy cofounder stake or real salary compensation. A strong engineer can earn well anywhere and has their choice of employer, so the levers of salary and equity have to be understood honestly rather than wished away.

Can I start a startup without a technical cofounder?

You can start, and you can get surprisingly far. No-code tools can produce a genuinely functioning prototype with just you and a designer, and that is sometimes enough to reach real revenue and prove the business. What you cannot do is stay there. If the business model actually works, you reach the point of needing an in-house technical cofounder faster than you expect, which is why the search has to run in parallel rather than after.

How do I vet a technical cofounder if I cannot code?

You outsource the judgement. Engineers vary enormously in ability and some will overstate what they know, claiming deep expertise in a framework they are reading up on that same night. I have always kept computer science PhDs in my life who will spend twenty minutes with a candidate at a whiteboard and tell me plainly whether the person is what they claim. There are also paid technical interview services. What you must not do is rely on your own read of a resume and a confident conversation.

Should a technical cofounder keep their job and work nights at first?

This is the single most common mistake I see, and it has a predictable failure curve. The founder is thrilled to have finally found someone who will put in two to four hours an evening. Roughly a month in, once the honeymoon ends, that person reports increased workload at their day job and disappears. I have watched founders cycle through three, four, five people this way inside six months, and the next engineer almost never wants to build on the previous one’s code.

How long does it take to find a technical cofounder?

Plan on three to six months to land someone in-house, which is exactly why you cannot start looking at the moment you need them. Founders typically discover the timeline when traction arrives and they suddenly need engineering capacity, by which point they are half a year behind. That said, urgency compresses it dramatically. I have seen a team that spent months failing to find anyone land a strong candidate in sixty hours once the consequences of not doing so became real.

Do I still need a technical cofounder if I can build with AI?

Eventually, yes, but probably not on day one, and that is a real change. AI coding tools can take a determined non-technical founder to a working product with paying customers, which was not realistic a few years ago. What they do not do is close the gap between you and a professional engineer, because the same tools made strong engineers dramatically faster too. There are also three things you cannot evaluate in your own AI-built product: whether the quality holds as it grows, whether it is secure, and whether it has any taste. Security is the dangerous one, because it fails silently.

Has AI made it easier to find a technical cofounder?

It has made it easier to recruit one on good terms, which is not the same thing. The supply did not improve. Half of strong engineers still will not join a startup, and AI arguably raised their market value. What changed is your side of the table. You can now arrive with a working product and revenue instead of a wireframe, and an engineer joining a company with customers takes a meaningfully smaller equity stake than one joining an idea. Vetting got harder, though, since anyone with a good agent produces impressive-looking code on a take-home exercise.

Do I need a technical cofounder before validating my idea?

No, and recruiting one first is usually a mistake. Before anyone writes a line of code you should have validated the problem, the customer, and the business model. Spending six months building something that any competent advisor could have told you was dead on arrival wastes the only resource a founder cannot replace, which is time. Validation also makes recruiting easier, because traction is the most persuasive thing you can put in front of a strong engineer.

---

John Richards is Managing Partner at Startup Ignition Ventures, a $20M pre-seed fund. He has made over 200 angel investments including Omniture, Fusion-io, Ancestry, Skullcandy, and Lyft, and led BYU’s entrepreneurship program for 12 years. This article is based on his conversation on recruiting a tech cofounder.