Conversation History

Grepp preserves your full conversation across every turn. This lets you ask natural follow-up questions, refine previous requests, and build on earlier results — all without repeating yourself.

How history works

Every message you send and every response the agent gives is stored as part of your conversation. When you send a new message, the agent has access to the full history of what was discussed, including:

  • Your previous questions and requests
  • Queries that were run and their results
  • Charts and dashboards that were created or updated
  • Any clarifications or refinements you made along the way

This means you can say things like "now break that down by region" or "actually, filter to just this year" and the agent understands exactly what you're referring to.

Long conversations

Grepp automatically manages conversation length so you never hit a hard wall. As conversations grow, the agent intelligently prioritizes the most recent and relevant context. Earlier parts of the conversation are still accessible — the agent can look back through history when you reference something from earlier in the session.

If you're working in a very long session, you can always ask the agent to recall something specific from earlier: "What was the query we used for the revenue chart?" The agent will look it up.

Session work

During your conversation, Grepp also tracks everything that was created or changed — charts, dashboards, queries — as a running log for the session. This lets the agent reference your work by name or ID without having to re-query or re-create anything.

For example, after you create a chart, you can say "update that chart to show monthly totals" and the agent already knows the chart's ID and current configuration.

Searching history

The agent can also search through your conversation when you ask it to look back:

  • "What did we query for the sales report?"
  • "Remind me what filters we applied earlier"
  • "What tables have we used in this conversation?"

See Recall for more on how the agent retrieves past work.