Claude
Agent Builder.

A custom GPT or OpenAI Assistant is a hosted product that runs inside their ecosystem with limited tool access, no real integration to your existing systems, and no eval framework. Claude Agent Builder produces an agent deployed into your environment, integrated with your business system, evaluated against your criteria, and owned by you in code form. If your use case fits inside a custom GPT's limits, you may not need us. If you want production-grade integration, eval, and ownership, this is the lower path.

We build slack into the schedule. If you miss a stakeholder review by 1-2 days, we absorb it. If you need to pause the engagement for more than 5 business days, we pause the clock — no extra fees — and resume when you're ready. If scope grows mid-engagement (you decide you want a second integration, or a new use case), we quote a change order and you decide whether to add to this engagement or scope a new one.

Highly use-case dependent. For most internal copilots and customer support agents at SMB volume (under 10,000 interactions/month), you'll typically see $50 to $400/month in Claude API costs. High-volume consumer-facing agents can run $500-$5,000/month. We model expected costs during week 1 based on your specific use case and projected traffic, and we build cost dashboards so you can monitor in real time. You pay Anthropic directly using your own API key — we never mark up token costs.

Almost certainly yes if it has a public REST or GraphQL API. Common integrations we've shipped at this tier: Slack, Microsoft Teams, Salesforce, HubSpot, Pipedrive, Zendesk, Intercom, Gmail, Outlook, Notion, Google Drive, Airtable, Linear, Asana, Jira, GitHub, Stripe, and many internal/custom APIs. For systems without modern APIs (some legacy ERPs, older SaaS tools), we'll tell you during the intake call whether it's feasible at this tier or whether it needs a custom Studios Growth scope.

Three reasons we default to Claude for agent work: (1) Claude's tool-use quality is consistently the strongest in production benchmarks we run; (2) Anthropic's enterprise API terms include strong "no training on your data" commitments by default; (3) Claude's instruction following at long context is notably better than alternatives, which matters when your agent has many tools or guardrails. Where Claude isn't the right choice for your use case (cost-sensitive high-volume consumer cases sometimes favor GPT; sovereignty cases favor open-source on your infra), we'll tell you during intake and route to Studios Growth where we have more model flexibility.

Three paths. (1) Your team operates it using the runbook and training video — many SMB clients do exactly this. (2) Pay-as-you-go: hourly support at $250/hr when you need it, no minimum. (3) Studios Retainer ($5K-$25K/month, 6-12 month minimum) if you want ongoing optimization, new features, or a dedicated engineer on standby. About 30% of Claude Agent Builder clients convert to retainer within 90 days of launch.

You own the custom code, custom prompts, and project-specific configurations developed for you — assigned to you on full payment. We retain ownership of our methodologies and reusable prompt-library patterns (our "Background IP"), but those are licensed to you perpetually as embedded in your deliverables. If you want to take the project to another vendor or your internal team six months from now, the repo is yours and there's no lock-in. Full IP terms in our Terms of Service §8.

The 50% deposit is non-refundable once we start week 1 work. If we materially fail to deliver against the eval rubric we agreed on in week 1, we'll work to fix it under our limited services warranty (re-perform or correct) — that's covered in the Terms §11. If you terminate without cause mid-engagement, you owe for work performed through termination. We're confident enough in fixed-scope productized work that we publish prices and timelines on the website — the warranty terms reflect that confidence.