Freelance developer, agency, or technical co-founder

Four ways to get a first product built, and the failure mode of each. A freelancer is cheap until they leave. An agency buys process and sells hours. A co-founder aligns incentives and costs equity. A build partner is the middle, and it suits fewer people than its marketing suggests.

You have four real options and everyone will tell you theirs is the right one, including me, so read this knowing where I sit. I run a build partnership. That is option four. I have also been options one, two and three at different points and I have watched all of them fail from close up.

Here is what each actually costs you, and not in dollars.

The short version

A freelance developer

Cheapest way to start and by far the most common. It works when the work is genuinely bounded. Build this screen, connect that API, fix this thing.

It stops working when the work needs judgment about what to build rather than how. A good freelancer will build what you ask for. If what you asked for was wrong, you will find out in month four, and the fix will be priced as new work because it is new work.

The real risk is not skill. It is continuity. One person with the whole system in their head is fine until they take another contract, get sick, or lose interest around week eight. Nothing in that list is a character flaw. It is just what one person is.

If you go this way, the thing worth paying for is documentation you can hand to someone else. Most people do not ask for it and it is the difference between a stalled product and a recoverable one.

An agency

You are buying process, capacity and someone to call. In environments where that matters, it is worth real money. Enterprise integrations, security review, a compliance track, five stakeholders who each need something different. That is a genuinely different problem and a sole developer will drown in it.

What you are also buying, usually, is distance. You talk to an account manager. Decisions are made in rooms you are not in. And the pricing model means the incentive runs against speed, because a three month program with fifty workshops earns more than doing the thing you actually needed in a week.

I know this because I ran the consulting version of it. We got good enough to spot the five minute fix and do that instead of the program. Happy clients. No money. In a service business, efficiency is a bug, and that is not cynicism, it is arithmetic.

Ask an agency what happens if it takes half as long as they think. The answer tells you which kind you are dealing with.

A technical co-founder

The incentives are perfect and everything else is hard.

Someone who owns a real slice of the company will care about the outcome the way you do. No scope negotiation, no hourly meter, no distance from the decisions. For a business that is genuinely a technology business, where the product is the company rather than a tool the company uses, this is usually correct.

The cost is control and it is permanent. Equity does not end when the build does. If you disagree about direction in year two, and you will, you are having that argument with a part-owner rather than a supplier.

There is also a supply problem nobody says out loud. Senior technical people who can build a product alone can also earn a great deal of money doing it for someone else, with no risk. The ones available to join an unfunded idea are, on average, either early in their career or between things. Some are excellent. The search is longer than people plan for, and the time spent searching is time not spent selling.

I do take equity in some things, so I am not against it. I would just say that the version where you give away thirty percent to get a first version built is usually a bad trade, and the version where you find someone after you have something working is a much better one.

A build partner

The middle option. Fixed scope, fixed date, fixed price, and it ends.

The case for it is that you keep the equity and the control, you get senior people for a defined window, and there is a date with money attached to it. Ours run six weeks in six stages, billed in sixths as each stage completes. Walk at any gate, pay nothing further, keep everything produced. The repository and every account sit in your name from day one.

The case against is that it only works if the scope is right at the start. If you do not yet know what you are building, no amount of process fixes that, and a fixed scope becomes a fixed way to build the wrong thing quickly. That is the honest failure mode and it is ours, not yours.

So the first stage is not building. It is deciding whether to. Sometimes the answer is that the thing you need is a spreadsheet and a Tuesday afternoon, and telling someone that costs me a build but saves them a year.

What actually decides it

Three questions, and none of them are about budget.

Is this a technology business, or a business that needs a piece of technology? If the product is the company, get a co-founder. If it is a tool that makes an existing business work, do not give away equity for it.

Do you know what you are building, specifically enough that someone could disagree with you? If yes, a bounded engagement works. If no, the honest first spend is on deciding, and you should be suspicious of anyone who quotes a build without that.

What happens if the person building it disappears in week five? Whatever answer makes you feel sick is the risk you are actually taking. Price that, not the hourly rate.

The one nobody picks

Build nothing yet.

Put up a page, put a real price on it, and see if anyone pays. You will learn more in a fortnight about whether this is real than three months of building will tell you, and if nobody pays you have saved the entire budget.

I have talked people out of builds this way and it is always the right call and it never feels like it at the time. The idea is not the asset. The proof that someone will pay for it is the asset. Everything after that is just execution risk, which is the kind you can hire for.

Common questions

How much equity does a technical co-founder take

Commonly somewhere between fifteen and forty percent depending on stage and how much is already built. The number matters less than what happens if it does not work out, so agree the vesting and the exit before the equity.

Can I just build it myself with AI tools

You can get further than you could three years ago, and much further than most technical people admit. Where it falls over is the last twenty percent, which is the part that was always hard. We have tested every tool going and the honest finding is that it works best late in a build, once the structure is already well defined.

What is the cheapest way to test an idea before building anything

Sell it first. A landing page, a real price and a real payment link will teach you more in a fortnight than three months of building, and if nobody pays you have saved the whole budget.