MCP use cases

What to do with Zentrik MCP

Once the connection works, the fastest wins are usually simple: read what the graph already says, add fresh evidence, or compare where to focus.

This page is for the moment after the connection works.

If you still need to add the server, start with Zentrik MCP setup.

The most reliable first moves are:

  • read the current signal before making a product call
  • turn fresh notes into a signal right after a conversation
  • compare opportunities using both ARR and evidence
Claude asking for approval before calling the Zentrik MCP signal creation tool.

Once connected, the client can call Zentrik MCP tools directly. This example shows a signal creation action from transcript text.

Start here

These prompts are intentionally short. Start here, then adapt them to the question in front of you.

Read the current signal

Use this when you want a fast read on what customers are saying before planning or prioritizing.

Prompt

Code3 lines
Pull all insights with provenance. Synthesize the top themes customers are
raising. Group by insight type, summarize each theme in 2-3 sentences, and
include one representative verbatim quote per theme.

Best for

  • weekly PM syncs
  • onboarding someone new to the problem space
  • getting an evidence-backed snapshot before a decision

Ask for quotes and provenance so the answer stays anchored to real evidence.

Capture fresh evidence

Use this right after a customer call, demo, or interview while the context is still fresh.

Prompt

Code6 lines
I just had a customer call. Here are my notes:

[PASTE NOTES OR TRANSCRIPT SNIPPET]

Create this as a Zentrik signal.
Context: [sales_discovery / user_interview / usability_testing / feature_feedback]

Why this matters

  • the notes enter the normal Zentrik signal pipeline immediately
  • extraction can happen from the same text
  • the evidence is available to the rest of the team, not trapped in your notes

Bullet-point notes work fine. A full transcript is helpful but not required.

Compare where to focus

Use this before roadmap, staffing, or prioritization conversations.

Prompt

Code4 lines
List all opportunities ranked by ARR (use computedArr). For each one, show:
name, status, computedArr, number of linked insights, number of linked ideas.
Add a brief evidence rating: Strong (3+ insights), Moderate (1-2 insights),
or Weak (0 insights).

Why this works

  • it combines revenue context with evidence strength
  • it surfaces which bets are both valuable and well-supported
  • it highlights high-ARR areas that still need more validation

What to do next

Treat the output as a working draft.

Good follow-up moves:

  • turn a synthesis into a team update, planning note, or next question
  • review a newly created signal in Zentrik after processing finishes
  • use ranked opportunities as an input to roadmap or initiative discussion
  • if you need to edit links, approve records, or clean up structure, go back to Zentrik

If you want a better answer, narrow the scope instead of adding more instructions. See MCP best practices.

Troubleshooting

These items are about workflow and expectations in the product, not a broken OAuth client. If something contradicts what you see in your workspace, note your workspace name and the screen, then contact us.

The answer is too generic

Ask for provenance, quotes, or a specific output format. Narrowing the scope by timeframe, topic, or product area usually improves quality more than adding more narrative instructions.

A newly created signal is not ready yet

Signal processing is asynchronous. Wait a short time, then ask the agent to fetch the signal again or review it in Zentrik once extraction has finished.

The workflow needs a different workspace

Reconnect the MCP client and choose the workspace you want. Zentrik MCP tokens are scoped to one workspace at a time.