Information, before any access
A clear description of the workflow, who the users are, and what data needs to persist — this drives the actual architecture decisions and matters more upfront than any credential.
Access typically needed
- Hosting/deployment access for the environment the application will run on
- API credentials for any third-party systems being integrated — see [Integrations](/ai/integrations/)
- A staging or test account for any CRM/tool being connected, if available
What's NOT needed by default
Financial/billing system access unrelated to the project, or admin access broader than the specific integration being built — see What Access Does an AI Automation Agency Need for the same principle applied to the AI-agent side of the business.
Who owns what afterward
Code, infrastructure accounts and data remain yours — nothing is built into an account controlled independently of you.
Questions
Do you need our production database credentials from day one?+
Usually not — a staging/test environment is used during build, with production access requested closer to launch.
What if we don't have clean documentation of our current process?+
That's normal — discovery during the Product Scope conversation is specifically meant to surface and document this, not assume it already exists.
Who owns the code afterward?+
You do — the codebase and infrastructure accounts remain under your control.