For agencies
You sold it. I build it under your name, and I never touch your client.
You run an agency. AI is on the services page because clients ask for it, and the last time one actually bought it, the AI layer was the part of the project you were least sure you could deliver on a fixed price.
That is a reasonable place to be. Auth, Stripe, CRUD and deploy are a solved problem for a good team. An extraction pipeline that has to stay accurate on real customer documents is a different discipline, and hiring for it takes four months.
I do that layer, under your brand, and I do not compete with you for anything else.
The promise
I do not sell the surrounding work, and I do not approach your client. Contractually.
This is the objection every agency has, so it is answered before you have to ask it.
The short answer
White-label AI development from Code and Proof means an agency sells the AI feature and Code and Proof builds it under the agency's brand. The arrangement is contractual: the surrounding work is never sold into the account and the end client is never approached directly. Priced by day-rate or fixed scope, depending on how defined the work is.
How it works
Your brand, not mine
Commits, documentation and handover are written to sit inside your repo and your process. Nothing I produce carries my name unless you want it to.
Client-facing or invisible
Your call. I can join calls as your AI engineer, or you can relay and I never appear. Both work; decide before the kickoff rather than during.
Scoped like your contracts
You quote fixed-price, so I scope fixed-price. The estimate you get from me is the one you can safely mark up, including the parts that usually blow up.
Handover is the deliverable
Your team maintains this after I go. Documentation, the eval harness and a walkthrough are part of the engagement, not an upsell.
NDA on request
Standard mutual NDA, signed before scoping. I have no interest in your client list.
I will not approach your client
In writing, in the contract. I do not sell the surrounding work, and I do not follow up after the engagement ends.
What I take on
Good fits:
- The AI layer of a fixed-price build you have already sold
- A feature that shipped, does not work well enough, and the client has noticed
- Scoping help before you quote — so the number you give is survivable
- An eval harness for something your team built and cannot measure
Not fits:
- General overflow development. I am not a body shop.
- Rescuing a project that is failing for reasons unrelated to the AI.
- Anything where you need me to pretend the deadline is achievable.
Rates are day-rate or fixed-scope depending on how well defined the work is. Ask and I will tell you on the first call rather than after three.
Common questions
- Will you approach our client directly?
- No, and it is contractual rather than a promise. On white-label work Code and Proof builds the AI feature under the agency's brand, does not sell the surrounding work into the account, and does not approach the end client.