Technical Co-Founder vs Fractional CTO vs Agency: Four Questions That Decide It (2026)

Short answer: four questions, asked in order, settle it. Is the code itself the thing you are selling? Do you need a permanent technical seat, or a finished product? What is the equity worth in money? And is anything built yet? A technical co-founder fits when the technical work is the company and will never be finished. A fractional CTO fits when you already have engineers and what is missing is someone senior to lead them. A senior product partner (an agency built for this job) fits when you need a launched product you own in the next few months and would rather pay cash than give up equity.

Most comparison pages list the three routes and leave you to weigh them. This one gives you the order to ask the questions in, a stop rule for each, and the questions to put to each option before you sign. It takes about ten minutes.

Last verified 2 October 2026.

The number that changed this decision

McKinsey's The state of AI in 2026, published in August 2026, found that 32 percent of respondents say their organisation decided against buying at least one software product or feature because it could be built internally with agentic coding tools (McKinsey, The state of AI, read 2 October 2026). Among the companies McKinsey calls AI high performers, it was nearly half.

That is a survey of established organisations, and it matters to a founder for its direction rather than its size. Building software is getting faster and cheaper relative to buying it. When a first version took a year and a team, you needed someone bound to you for a decade to see it through, and equity was the only currency that bought that commitment. That is where the even co-founder split came from. The build is no longer that slow, which is why the three routes below are now real alternatives rather than one option and two compromises.

Question 1: is the code the thing you are selling?

Ask it precisely. Not "is my product software", because almost every product is. Ask whether the hard technical work is the defensible asset: a new model, a research result, an algorithm a competitor cannot reproduce.

• If yes: the technical work is the company, and the person doing it should own part of it for the long run. Look for a technical co-founder, and run the seven checks before you grant equity.
• If no, keep going. If the product is built on well understood foundations and your edge is customers, domain knowledge, distribution or operations, then the code is the vehicle and not the cargo. Most software-enabled businesses sit here, and many talk themselves into the first category by mistake.

Question 2: do you need a permanent seat, or a finished product?

These are different needs and they are often confused.

• A permanent seat is someone who owns technical decisions indefinitely, hires and manages engineers, joins board conversations and carries the technical risk of the company. That is a co-founder or a CTO.
• A finished product is a working, launched first version, built to a written scope, on a date, by people who have done it before. That is a delivery relationship.

A fractional CTO is the permanent seat, rented part time. It is a strong answer when you already have engineers who need direction. When there are no engineers yet, a fractional CTO's first job is to hire them, so you pay leadership rates for a recruitment project and still have nothing built. What fractional CTOs publish, and the advice-plus-build total, is on fractional CTO cost in 2026.

• Stop rule: zero engineers and no product means a fractional CTO is early. Revisit it once something exists and someone needs to lead its growth.

Question 3: what is the equity worth in money?

This is the question founders skip, and it is the only one with arithmetic in it.

Equity feels free because no money leaves your account. Convert it before you decide. Y Combinator's Startup Library advises proposing an even split to a technical co-founder when you cannot pay a salary (YC, How to Find a Technical Co-Founder, read 2 October 2026). If the company is one day worth $10 million, half of it is $5 million. If it is worth $1 million, half is $500,000. Put that next to a cash cost you pay once. How much equity a technical co-founder gets covers how the number moves with timing and risk.

For the cash side, KUMO publishes three bands and only three: a Starter Build of a first production version at $15K to $50K, a larger multi-system product at $50K to $100K, and ongoing engineering at $5K to $10K per month. The AI development cost calculator places your own scope in a band.

• Stop rule: if you can fund the cash cost of the first version, and that cost is lower than what you would honestly value the equity at, the equity is the more expensive option. Say both numbers out loud before you sign anything.

Question 4: is anything built yet?

The order of operations changed and most advice has not caught up. You can now build a working prototype yourself with AI tools before you commit to any route. The prototype tells you whether real users care, turns a vague brief into something an engineer can price, and makes you far more credible to a technical co-founder if you still want one.

• Stop rule: if nothing is built, build the prototype first, whichever route you pick later. Every route gets cheaper and better specified once one exists. The walkthrough is in build a prototype with AI tools.

The three routes in the same units

Comparison pages usually mix units, quoting equity for one route and hourly rates for another. Here they are on the same rows.

How to read these numbers. Subscription prices are the vendors' published list prices on the date shown, before taxes, overages, add-ons and seat growth. The custom-build figures are KUMO's published bands for the build only: hosting, model usage, third-party fees and ongoing engineering ($5,000 to $10,000 per month if you have no engineers) sit on top, and we say so wherever a total is compared. A standard tool is the right answer when your workflow fits it; configuration is enough when the gap is small; a custom build is justified only by a specific complexity named on this page, not by preference.

The row almost nobody publishes is ownership, and it is the one that matters most if the relationship ends. Technical co-founder left: who owns the code shows what happens when it was never written down.

Questions to ask each option before you sign

Whichever column you land in, ask these in writing and keep the answers.

A prospective technical co-founder

1. How many hours a week, starting when, and what would make you leave?
2. Will you accept four-year vesting with a one-year cliff, and assign all code and IP to the company?
3. Can we do a paid two-week trial on a real piece of the product first?

A fractional CTO

1. How many hours a month, and which decisions are yours versus mine?
2. Who writes the code: you, engineers you hire, or engineers I hire?
3. What does the handover look like if I hire a full-time CTO next year?

A senior product partner

1. What is in and out of scope, in writing, and what does a change cost?
2. Who owns the code, the cloud accounts and the domain, and from when?
3. Who are the actual engineers on my build, and what have they shipped and run themselves?

If an answer is vague, that vagueness is the answer.

Where KUMO sits

KUMO is the third column. The founders built Volopay's first production version as its founding engineers, and Volopay went on through Y Combinator and raised institutional funding: the technical co-founder job, done as a delivery partner. KUMO also runs CampaignHQ, its own product, in production since 2022, so the team operates software it owns as well as software it builds for others. The Volopay proof is direct: KUMO's founders did the founding-engineer job this article compares, as a delivery partner.

The engagement is set out on the founders partnership page: a fixed scope against the published bands, senior engineers, milestone delivery, two to three months to launch, and a clean handover to your own CTO whenever you hire one. You keep your equity and own the code from day one. The full Build-First Path is on the technical co-founder alternative page.

Run the four questions in ten minutes

1. Is the code the defensible asset? If yes, find a co-founder and vet them properly.
2. Permanent seat or finished product? No engineers yet means a fractional CTO is early.
3. Convert the equity into money and compare it with a cash number that has a scope attached.
4. Build the prototype first either way.

If you land on the third column, book a free 30-minute call and bring whatever you have: a prototype, a document, or just the problem. You will leave with a scope and a band, not a pitch.

FAQ

Should I get a technical co-founder, a fractional CTO or an agency? Ask the four questions above in order. A co-founder fits when the technical work is the company. A fractional CTO fits when you have engineers who need leadership. A senior product partner fits when you need a launched, owned product soon and prefer cash to equity.

Who should a non-technical founder hire to build their AI product? If the AI work applies existing models to your domain and you want a launched product rather than a permanent technical partner, a senior product partner is the shortest path. If the model itself is the company, look for a technical co-founder.

Is a fractional CTO cheaper than an agency? They buy different things. A fractional CTO buys leadership hours; a delivery partner buys a finished product against a scope. With no engineers to lead, a fractional CTO costs leadership rates plus the team they then have to hire.

How much equity does a technical co-founder get? YC advises an even split when you cannot pay a salary. The number moves with when they join, what they give up, how much cash you can pay and how much risk is already retired.

Do I own the code if I use a delivery partner? With KUMO, yes, from day one. Ask any partner this in writing before you sign.

Can I change route later? Yes, and it is common. Founders often build a first version with a partner, reach revenue, then hire a CTO to take it forward. Because you own the code, that is a normal engineering handover rather than a negotiation.

Last verified 2 October 2026. McKinsey figure from The state of AI in 2026 (August 2026); YC guidance from the YC Startup Library; both read on that date. KUMO bands are KUMO's published prices.