Anthropic Claude Startups Programme: What It Is, How to Apply

According to the programme page, every approved company receives two core things: $1,000 in Claude API credits and one year of free Claude Team. The Team offer covers up to five Premium seats and applies only to organisations that are new to Team.

A few extras sit alongside. Anthropic's Applied AI team runs live office hours every two weeks, in 45-minute sessions. Members who receive credits get higher API rate limits. Startups backed by a venture capital fund in the partner network can receive up to $100,000 in additional API credits through that fund.

There is also the Claude Startup Stack: offers from independent third-party companies, with a combined value of up to $45,000. These offers belong to those companies, not to Anthropic, and the codes are redeemed inside Claude Console. Each offer therefore has its own terms, and you should read them separately.

• $1,000 in Claude API credits, which expire six months after they are granted.
• One year of free Claude Team, up to five Premium seats, only for organisations new to Team.
• Up to $100,000 in additional API credits through a venture capital fund in the partner network, for startups that fund backs.
• The Claude Startup Stack of third-party offers, with a combined value of up to $45,000.
• Office hours every two weeks, in 45-minute sessions, and higher API rate limits.

The eligibility test is broad. Startups founded in the last 5 years, or funded in the last 2 years, can apply. The programme is open to bootstrapped companies, pre-seed startups and funded ones.

The application asks for three things: a Claude Console account, a company email address that matches your website's domain, and a short description of what you do. You must also comply with Anthropic's supportability policies.

The company email requirement looks minor, but it is the most likely place to stumble. If email on your domain does not work properly, or the founders still deal with everything from personal addresses, the application slows down. The free email DNS check on our site shows whether the basic records for your domain's mail are in place before you apply.

You apply by signing in to Claude Console and completing the form. According to the programme page, most decisions arrive within minutes, and applications that need manual review usually take 2 to 3 business days. Existing members do not need to reapply.

A little preparation before you open the form speeds things up.

• A company email address on your own domain that is actually monitored.
• A Console account created with that same address.
• A homepage that says plainly what you do, so your description and your site agree.
• A quick review of whether your product fits Anthropic's supportability policies.
• A short note on which product feature the credits will go to; the form may not ask, but it sharpens your plan.

The credits apply only to the first-party Claude API used through Claude Console. They cannot be used on AWS Bedrock, Google Cloud Vertex AI or any other third-party platform. If your product calls models through one of those platforms today, the credits will not cover that traffic.

In practice, depending on Console means that to spend the credits you must connect the relevant part of your product directly to Anthropic's API.

Time is another dependency. The credits end after six months, and after that you pay.

Because the window is limited, it is healthiest to treat the credits as an experiment budget. The first task is to write down the question the credits will answer: does this feature change something for our customers, do users come back to it, does the cost fit inside our pricing?

Then split the work into small pieces. Put one use case into production first, watch real usage, and only then move to a second. Keep a simple record of what each call costs from day one. The measurements you take in the first weeks become the basis for the pricing decision you make in month six.

Use the office hours with a concrete question.

• Write down the one question the credits should answer.
• Take one use case live, measure it, then widen.
• Record spend per call from the first day.
• Work out the monthly cost after the credits end a month early and compare it with your product price.
• Go to office hours with a concrete question.

The first mistake is assuming the credits work on every platform. For a product that runs through Bedrock or Vertex AI, they are of no use. The second is applying with a personal email address, or not checking that mail on your domain works.

The third is spending the credits on experiments without measuring anything. If at the end of six months you hold no data on what worked, the credits were simply a gift used once.

Finally, do not rely on memory for the terms and dates. For questions such as how credits are treated for tax or accounting, read the terms on the programme page and consult your accountant.

If your company was founded more than 5 years ago and has not been funded in the last 2 years, you may not meet the eligibility test.

If your product has no link to artificial intelligence yet, do not invent one just because credits exist.

The deciding question is simple: is there already a concrete job in my product plan that these credits would pay for? If yes, the programme is a good lever. If not, clarify the plan first and apply afterwards.