Knownbase Client Onboarding Sprint (5-7 days)
A structured onboarding playbook that configures Knownbase for a client's AI coding agents, establishes note taxonomy, and trains the team to maintain persistent project memory across sessions. Time: 5-7 days.
By InnovaAI ResearchPublished
Knownbase Client Onboarding Sprint (5-7 days)
A structured onboarding playbook that configures Knownbase for a client's AI coding agents, establishes note taxonomy, and trains the team to maintain persistent project memory across sessions.
- Client has an active Knownbase account (Free or Go plan) and at least one AI coding agent (Claude Code, Cursor, Codex, or MCP-compatible) in use.
- Access to the client's code repositories and permission to install MCP servers in their development environment.
- A defined list of client projects (up to 10 for the Go plan) that will be connected to Knownbase.
- Client team availability for a 1-hour training session on day 5 or 6.
- Agency-side admin credentials to create workspaces and manage MCP keys.
- 1.Create a Knownbase account and set up the MCP endpoint for the client's primary AI coding tool (e.g., Claude Code).
- 2.Configure the MCP keys for the client's agent, ensuring secure access to the Knownbase server.
- 3.Test note creation, retrieval, and revision history to validate basic functionality.
- 1.Create project workspaces in Knownbase for each client codebase (up to 10 on the Go plan).
- 2.Define tags and statuses for notes (e.g., 'architecture', 'debugging', 'constraint', 'open', 'resolved').
- 3.Set up webhooks for real-time notifications on note changes.
- 1.Integrate the client's AI agents (Cursor, Codex, etc.) with Knownbase via MCP, ensuring all agents can read and write notes.
- 2.Document the note-entry workflow: agents write notes on decisions, debugging discoveries, and constraints.
- 3.Verify that shared workspaces are accessible to all client team members.
- 1.Build a structured note taxonomy for architectural decisions and debugging logs, tailored to the client's projects.
- 2.Create template notes for common scenarios (e.g., 'ADR: [Decision]', 'Bug: [Issue] - [Resolution]').
- 3.Test search queries to ensure relevant notes are retrievable on demand.
- 1.Train the client team on search workflows and note-entry best practices (1-hour session).
- 2.Provide a quick-reference guide for using Knownbase with their agents.
- 3.Review any questions and adjust the taxonomy based on feedback.
- 1.Monitor agent interactions with Knownbase for 24 hours to identify any context-loss issues.
- 2.Refine note tags and statuses based on real usage patterns.
- 3.Document any troubleshooting steps for common MCP connection issues.
- 1.Deliver the onboarding guide and finalize the configuration.
- 2.Conduct a wrap-up meeting to confirm the client team can independently maintain project memory.
- 3.Hand over admin credentials and provide ongoing support contact.
The agency charges a $1,800 setup fee for roughly 16 hours of work, yielding an effective hourly rate of $112.50. The client's ongoing Knownbase subscription is only $2.50-$5 per month, so the agency can bundle this cost into a monthly retainer (e.g., $50-$100/month) for maintenance and support, generating high margin with minimal overhead.
- Configured Knownbase workspace with MCP keys for client's primary AI coding tool
- Structured note taxonomy for architectural decisions and debugging logs
- Documented onboarding guide for client team to maintain and query project memory
- Training session recording and quick-reference guide
- Webhook and notification setup for real-time updates
The client's AI agents can successfully write and retrieve notes in Knownbase across sessions without manual context-setting, and the team has completed training and can independently manage their project memory.