Claude Integration
Claude + Diaphora
Use Diaphora as the structured workflow engine and Claude as the collaborative surface for refining the final narrative.
Overview
How the Claude workflow fits
The Claude integration pattern is about handoff. Diaphora gathers, structures, and renders the result; Claude becomes the conversational layer teams use to adapt that output into an email, memo, strategy note, or stakeholder brief.
Knowledge workers run complex enterprise AI workflows directly through Diaphora’s MCP server. Diaphora handles the deterministic heavy lifting — gathering data, enforcing structure, and producing reliable first-pass outputs. Claude then focuses on the high-value creative work: refining tone, brainstorming ideas, and adapting the content for the specific audience.
Instead of asking Claude to handle the entire workflow from scratch, Enterprises build Diaphora plans for their Knowledge works to run deterministic, reliable AI powered workflows work through its MCP server. Claude is reserved only for the high-value creative final step: editorial refinement, brainstorming, and audience adaptation. The result: dramatically fewer prompts and significantly lower token spend.
Setup
How to integrate Claude with Diaphora
Diaphora is exposed as an MCP server. The setup is intentionally lightweight: connect your Diaphora account once, then add Diaphora as a custom connector inside Claude.
1. Start with a Diaphora account
Make sure you have an active Diaphora account before configuring Claude. That account is the identity layer Claude will connect through.
2. Open Claude and add a custom connector
Inside Claude, add Diaphora as a custom MCP connector. This points Claude at Diaphora's MCP surface instead of treating the workflow as a pasted prompt.
3. Complete the Diaphora IAM login flow
When Claude initiates the connection, you will be redirected into the Diaphora IAM login process. Authenticate there to authorize the connector.
4. Return to Claude and start using Diaphora tools
Once the IAM flow is complete, Claude can access Diaphora as an MCP server and use connected plans and workflows as part of your working session.
How it works
Working with the integration
1. Run the deterministic workflow in Diaphora
Start in Diaphora when the job needs connector access, repeatable extraction, or schema-shaped output.
- Diaphora gathers the operational context from the right tools.
- The plan enforces structure before any final narrative layer is added.
2. Use the rendered document as your base draft
If the plan returns a document, that document is the ideal starting point for Claude. It already captures the core facts, structure, and recommended framing.
- This reduces prompt thrash and cuts down on missing context.
- The team starts from a finished artifact instead of a blank box.
3. Ask Claude to adapt, not reconstruct
Claude is best used to translate the Diaphora output for a specific audience or purpose: a board memo, customer email, strategy note, or internal summary.
- Ask for style shifts, condensation, expansion, or alternate framings.
- Keep the factual backbone anchored to the original Diaphora output.
4. Keep Diaphora as the source of truth
When the underlying facts change, rerun the plan in Diaphora rather than manually rebuilding the context in Claude.
- This preserves repeatability across every refresh.
- Claude stays focused on collaboration instead of data wrangling.