Launch plan

Applications

Where to ask OpenAI and Anthropic for approval, with a draft for each form.

Where to apply

Neither is a license you buy.

CompanyWhereWhat it is
OpenAISign in with ChatGPT interest formA waitlist. OpenAI's guide says paid, commercial or remotely hosted apps should join it before offering the sign-in to users
AnthropicContact sales formNo application exists. This is where Anthropic's legal page sends questions about permitted sign-in methods

For Anthropic you also agree to the Commercial Terms of Service. How those are formally accepted is not confirmed. It is likely when you create a Claude Console account.

For OpenAI you can build and test before approval with the quickstart and the development kit.

Before you send

Draft for OpenAI

Product: Ad Shots, a desktop app for macOS and Windows that creates Meta ads for
e-commerce brands. It renders the customer's real product in 3D and an AI agent
researches ad angles and builds editable ad layouts in feed, story and square sizes.

Company: [company name], [website]
Stage: in development, not yet launched. Planned launch: [date].
Pricing: paid subscription, $99 a month.

What we are asking for: approval to offer Sign in with ChatGPT with plan usage, so
customers can run the app's AI agent on their own ChatGPT plan instead of an API key.

How we would use it:
- The app runs locally on the customer's computer. Sign-in completes through
  OpenAI's own flow using the official development kit. We do not collect or store
  ChatGPT credentials.
- Each customer uses only their own plan. We do not pool, share or resell usage.
- Usage is interactive: one person making ads, with the agent working on a request
  they typed. No background or unattended batch jobs.
- We check for the plan-usage permission before enabling AI requests, and keep
  "use your own API key" as an alternative.

Our subscription fee covers the app, the 3D rendering and the customer's library.
It is not a charge for model access.

Expected volume: [your estimate, e.g. under 500 customers in the first year].
Contact: [name], [email]

Draft for Anthropic

Subject: Confirming a permitted use of Claude Code inside a paid desktop app

We are building Ad Shots, a desktop app for macOS and Windows that creates Meta ads
for e-commerce brands. An AI agent in the app researches ad angles and builds
editable ad layouts. It is a paid subscription, in development and not yet launched.

We would like customers to be able to run that agent on their own Claude Pro or Max
plan. Having read the "Authentication and credential use" and "Can customers offer
Claude Code in their products?" sections of the Claude Code legal page, our planned
design is:

- Ad Shots runs the unmodified Claude Code program on the customer's own computer.
- The customer signs in to Claude Code themselves, through Anthropic's own sign-in
  flow. We do not offer Claude login in our interface, and we never collect, store
  or pass on Claude credentials or session tokens.
- Each customer's usage is billed to them by Anthropic under their own plan. We do
  not pay for, resell or intermediate Claude usage. Our fee covers the app itself.
- Customers can use their own API key instead.
- Usage is ordinary individual use: one person at their desk making ads.

Questions:
1. Is this design permitted for a paid product?
2. Do we need to do anything beyond agreeing to the Commercial Terms of Service?
3. Are there limits on how we describe this to customers? We plan to say, in plain
   text, that Ad Shots runs Claude Code.

Company: [company name], [website]
Contact: [name], [email]

Grok

There is no form. Read xAI's three legal pages in a browser first, then ask xAI whether a paid desktop app may drive Grok Build on the customer's own subscription. Links are on Sources.

Updated 2026-11-14. Built 2026-11-15.