Croft Client Workspace Setup (Onboarding)

Croft Client Workspace Setup (Onboarding)

1. 01

   Confirm the client's app was built by chatting with Claude, ChatGPT, or Cursor

   Croft only hosts apps on its own runtime, so an app built elsewhere cannot be migrated into the private server. Ask which assistant produced the app and whether the code is ready to deploy before quoting any retainer work.

2. 02

   Create the client workspace on the Solo tier at $24/month or the Team tier at $99/month

   Solo covers up to 10 people with shared data and automatic backups. Team raises the seat ceiling to 50 and adds roles and permissions for admins, creators, and members. Pick Team when the client's own staff will log in, not just your delivery team.

3. 03

   Deploy the app to the client's private server and verify zero-downtime deploy completed

   Automatic backups and zero-downtime deploys are included on both tiers. Run one deploy, then one redeploy, to confirm the client's app survives a release without a maintenance window.

4. 04

   Wire the external connectors the app depends on

   Croft ships connectors for Stripe, HubSpot, and 45+ other tools. Connect only what the app calls at runtime, then confirm shared data flows across the client's apps if they run more than one.

5. 05

   Connect the client's AI assistant to Croft over MCP

   The MCP tools expose errors, metrics, and logs to Claude or ChatGPT so the assistant can read failures and propose fixes. Test the connection by asking the assistant to pull the latest error group for the deployed app.

6. 06

   Configure error grouping, log alerts, and uptime health checks

   Croft auto-groups errors with stack traces and tracks slow SQL, response-time breakdowns, throughput, and failure rates. Set alert emails to route to your delivery inbox first, then to the client once the first week is stable.

7. 07

   Set the log retention expectation with the client in writing

   Searchable logs cover 7 days. Anything older is gone, so agree on what gets exported or archived elsewhere if the client has a compliance or postmortem requirement beyond that window.

8. 08

   Hand off the Croft dashboard and document the escalation path

   Train the client on reading error groups and response-time charts, then define who responds to an alert email and within what window. Retainer scope should state that monitoring escalation sits with your agency unless the client opts into self-service.