A domain founder without a technical co-founder usually has three options. They can raise money and hire, which is slow and assumes investors will back a company with no product. They can pay an agency, which produces code but leaves no one responsible for the technical decisions. Or they can work with a venture studio (some call it a lab), which invests capital and builds the product alongside the founder.
The third option has grown quickly, and so has the range of terms attached to it. One industry network, GSSN, reports that studio-built companies reach Series A faster than other startups. The most-cited industry figure is about 25 months against 56. That data is self-reported, and it comes with heavy selection bias. The more useful question is not whether studios work. It is whether this studio's deal works for your company.
1. How much equity, and what does it pay for?
Studio equity ranges widely. Commentators cite anywhere from 15% to 50%. The largest independent study we have seen maps 1,107 studios worldwide; across 38 pre-seed deals from 23 of them, it puts the median studio stake at 17%, with the middle half of deals between 12% and 23%. The stake matters most two rounds from now. One published model shows the difference. A studio that takes 25% at incorporation leaves the founder with about 48% after Series A. At 40%, the founder is left with about 38%.
Ask the studio to separate the equity it takes for capital from the equity it takes for work. If both are bundled into one number, you cannot tell whether the price is fair.
2. Who owns the code, the models and the IP?
This is the question founders forget and investors ask. Code written by a studio's engineers is not automatically your company's property. Without present-tense assignment language in the contract, ownership can stay with the studio, or with whichever contractor wrote the code. Due diligence at Series A will find this.
Ask for three things: an assignment of IP to the company as it is created; repositories held in the company's own organisation from day one; and clarity on any studio components that are licensed to the company rather than owned by it.
3. How is the engineering paid for?
Some studios charge fees on top of equity. Some pay for the build from the seed capital they invest. Some mix the two. None of these is wrong, but paying twice for the same work is. Some well-regarded studios, such as Alder VC, publish their terms before the first meeting. Some have removed portfolio fees altogether.
4. What happens when you hire your own CTO?
A studio should make itself unnecessary for engineering over time. Ask what a handover looks like: documentation, architecture decisions that are written down, a codebase a new CTO can read, and a timeline tied to the company's own hiring plan. One of the criticisms founders raise most often about studios is dependency: the product works, but only the studio's engineers can change it.
5. Who carries the operating load after launch?
Launch is where most agency relationships end and most production problems begin: security patches, backups, monitoring, incident response and model drift. For an AI product, the operating load also includes evaluation. Someone has to notice when the model's outputs change. Ask who does this, how it is paid for, and what happens if the studio relationship ends.
A good studio deal is one you could explain to your Series A investor in one page without apologising for any of it.
How wGrow Labs answers
We invest up to SGD 500K in seed capital for selected ventures, released against milestones. We take a minority stake. Engineering comes from wGrow Technologies, which has run production systems since 2008 and stays involved after launch. We build towards a handover to the company's own team.
- Inniches, venture studio research (2024): 1,107 studios mapped; equity data from 38 pre-seed deals at 23 studios (PitchBook)
- Alder VC, venture studio equity and dilution model
- GSSN data via Bundl: time to Series A (self-reported)
- Hylton Rodic, venture studio IP risk
- Venture Studio Forum, common studio failure modes