“What Are You Building on Claude?” — 5 Copy-Paste Templates for Your Startups Application
How to Use These Templates
Each template below covers both required application fields: (1) "What are you building on Claude?" and (2) "Where do you want support from Anthropic?" — both capped at ~500 characters. Customize the bracketed placeholders with your actual product name, target customer type, and specific Claude model. Do not copy verbatim — reviewers recognize repeated boilerplate.
What Makes a High-Scoring Answer
Anthropic's reviewers read thousands of applications. They are specifically looking for four signals of genuine product-level integration:
5 Ready-to-Customize Templates
Other Application Fields: Quick Guide
Select 21–40% if you are splitting inference between Claude and other providers (e.g. Gemini Flash for cheap routing, Claude Opus for complex tasks). Select 60%+ if Claude is your primary provider. Higher Anthropic spend percentages signal stronger commitment and may receive higher credit grants.
Select API / SDK if you call the Anthropic API directly from code. Select Claude Code if you use the Claude Code CLI for developer tooling. You can select multiple options. If you use Claude Team workspace AND the API, check both.
Frequently Asked Questions (FAQ)
Q:How many characters should my 'What are you building on Claude?' answer be?
Anthropic's application interface sets a character limit of approximately 500 characters for this field. Your answer should be substantive enough to fill close to that limit without being padded. Reviewers respond better to technically specific, dense answers than to short vague ones — aim for 450–500 characters.
Q:Can I copy these templates exactly, or should I customize them?
You should customize the templates with your actual product name, target customer segment, and specific Claude integration details. These templates give you the proven structure and technical vocabulary that passes reviewer checks. Replace bracketed placeholders ([Product Name], [Your Industry]) with real specifics — reviewers can recognize generic boilerplate.
Q:What should I write in 'Where do you want support from Anthropic?'
The 'support' field is your chance to tell Anthropic what you need beyond credits. Strong answers request specific things: rate limit tier elevation for high-volume pipelines, early access to beta model releases, technical office hours with Applied AI engineers, or guidance on prompt caching architecture. Vague answers like 'general support' score poorly.
Q:Does mentioning Claude Opus 5.5 vs Sonnet vs Haiku affect approval?
Yes. Mentioning specific model names signals technical depth and intentionality. If you explain why you use Claude Opus 5.5 specifically (e.g. its 200K context window for document analysis, high-reasoning quality for multi-step agents), reviewers understand you have done your homework. Generic 'Claude' references without model specificity read as superficial.
Q:Should I mention competitor models (GPT-6, Gemini) in the application?
Briefly mentioning why you chose Claude over alternatives is a positive signal. For example: 'Claude Opus 5.5's 200K context window and superior instruction-following outperformed GPT-6 Sol in our internal evals for long-document legal analysis.' This shows you evaluated alternatives and made an informed architectural decision.