Ship the ChatGPT Connector your customers will actually use
Satsuma helps companies launch branded connectors for ChatGPT, Claude, Gemini, Perplexity, Copilot, and other AI surfaces. We handle the connector plumbing: MCP server scaffolding, app UI, auth flows, hosted deployment, analytics, and observability.Public endpoint documentation is intentionally not published yet. These docs explain how a Satsuma connector launch works, what your team provides, and what Satsuma owns.
Connector Strategy
Map high-value assistant actions to the systems, permissions, and user journeys your product already supports.
MCP Buildout
Scaffold a production MCP server and companion app from your existing API and brand requirements.
Launch Operations
Deploy, monitor, analyze, and iterate on connector usage after launch.
What Satsuma Builds
MCP Server
A connector backend that exposes the right product actions to AI assistants with clear tool descriptions, scoped permissions, and platform-specific compatibility.
MCP App
Branded UI surfaces for confirmations, previews, settings, account linking, and moments where the assistant needs user-visible context.
Authentication
Account linking and authorization flows that keep user consent explicit while aligning with your existing identity and permission model.
Hosting & Deployment
Production deployment, environment configuration, release support, and channel-specific submission guidance.
Analytics
Usage instrumentation across sessions, tools, intents, errors, conversions, and customer-visible outcomes.
Observability
Logs, traces, health checks, and triage workflows so your team can understand what happened when an assistant action succeeds or fails.
Launch Workflow
1
Identify the assistant use cases
We work with your team to choose the workflows customers are already trying to complete through AI assistants, then define the actions, data, and permissions required.
2
Connect to your existing systems
Your team provides API access, sandbox credentials, brand guidance, and business rules. Satsuma turns those inputs into a connector-ready MCP implementation.
3
Design consent and account linking
We define the user-facing auth experience, required scopes, token handling, and revocation behavior before launch.
4
Deploy and validate
Satsuma hosts the connector, runs acceptance checks across target AI channels, and validates logging, analytics, and failure behavior.
5
Launch and improve
After launch, we monitor real sessions, surface usage insights, and help tune tools, prompts, and UI states based on what customers actually do.
What Your Team Provides
System Access
Existing APIs, staging environments, auth documentation, test accounts, and any operational constraints Satsuma should respect.
Product Intent
The customer workflows you want assistants to support, plus the success criteria that determine whether the connector is useful.
Brand Requirements
Logo assets, tone guidelines, legal language, and UI expectations for any customer-visible connector surfaces.
Launch Review
Stakeholder approvals for auth, data handling, security posture, and final behavior in the target AI channels.
Target Channels
Satsuma can help you launch across the assistant channels your customers use:ChatGPT
MCP apps and connector flows for ChatGPT experiences.
Claude
MCP-compatible tools and resources for Claude workflows.
Gemini
Connector patterns that adapt your capabilities to Gemini user journeys.
Perplexity
Research and answer flows that can use your authorized product context.
Copilot
Enterprise assistant workflows that fit Microsoft environments.
Custom MCP
Private or partner-specific MCP deployments for controlled launches.
Read Next
Identity & Authorization
Plan account linking, permission scopes, and token handling.
Connector Surfaces
Understand the MCP server, app UI, and assistant-channel surfaces Satsuma ships.
Observability & Support
See how logs, traces, analytics, and support workflows fit together.
Launch Plan
Review the milestones from kickoff through production rollout.