Set up Zentrik in ChatGPT

Use Zentrik in ChatGPT when you want to search customer evidence, inspect the source trail behind an opportunity, or capture a new customer note without rebuilding the context in another tool.

Zentrik organizes and returns the records. Your team still decides what to build.

Installation takes a few minutes. You need access to a Zentrik workspace and a ChatGPT or Codex surface that supports plugins.

Begin with a question your team is already trying to answer. Ask for the source records when the answer could affect a product decision.

Use the installed plugin to work with customer evidence:

• Search customer evidence, findings, opportunities, and ideas.
• Fetch the complete context for a search result.
• Summarize repeated customer problems across recent evidence.
• Review the evidence and linked findings behind an opportunity.
• Capture a new customer note as evidence.
• Check whether newly captured evidence has finished processing.

You can also plan product work and review Studies:

• Compare product options in an interactive decision brief, with linked evidence and open questions.
• List, answer, and revise an Idea's planning questions.
• Regenerate an existing Initiative Brief, PRD, or TDD from current Initiative state and check the regeneration job.
• Delete one Idea comment when the connected user is its author or a workspace owner/admin.
• List and inspect Studies, read aggregate evidence and authorized response details, create or refine an unpublished draft, and generate draft findings for human review.

OpenAI reviews new and changed tool definitions through continuous review. A new tool becomes available after its checks pass; an updated tool keeps its previous definition until the update passes. Your available tools also depend on workspace permissions and the access approved for your connection. Zentrik remains the place to publish Studies, invite participants, and manage the broader product graph.

Explicit destructive actions

• Deleting an Idea comment permanently removes the single addressed comment.
• Regenerating an Initiative Brief, PRD, or TDD replaces its current generated content.
• Answering an Idea question replaces its saved answer when one exists.
• Updating a Study draft replaces the supplied editable plan fields. It requires the latest draft revision and cannot change a Study that has responses.

Identify the exact target and confirm the replacement or deletion before acting. Zentrik checks permissions on the server.

• Workspace-scoped: The connection uses the single workspace selected during authorization.
• User-scoped: Zentrik applies the connected user's workspace access.
• Evidence capture: The plugin can capture evidence after you ask it to save supplied source material.
• Planning and Study writes: These tools answer Idea planning questions, regenerate generated Initiative documents, and create or refine unpublished Study drafts. They do not publish a Study or invite participants.
• Response access: Study response details can include answers, interview notes, transcripts, and known participant identity. Access is limited to the connected user's authorized workspace and requires the dedicated response scope.
• Destructive actions: Comment deletion, document regeneration, saved-answer replacement, and Study draft replacement affect only the identified target. Confirm the change before acting.
• Durable records: Evidence, findings, and opportunities stay in Zentrik after the ChatGPT conversation ends.

You can review Zentrik's privacy policy, security practices, and sub-processors before connecting a workspace.