Skip to main content
B2B · Product · For In-app product and AI agent owners · Also for B2C companies In-app Chat Insights reads every conversation users have with the AI agents in your product, from assistants that answer questions to agents that take actions for them. It surfaces the missing features, bugs, knowledge gaps and improvement opportunities in those chats, and drafts the tickets and docs to act on them.

What it reads

Conversations from the AI agents in your product, your knowledge base, support conversations and your Linear backlog. Usage and billing come through your warehouse, the Bulk API or a CSV, and a Jira backlog through a CSV.

How it works

  1. Ingest. Every chat with the AI agent in your product.
  2. Analyze. The gap behind every request it couldn’t handle.
  3. Act. Tickets and docs drafted before the gap grows.

What you get

Missing features, bugs and knowledge gaps, with tickets and docs drafted.
  • AI agent digest (Product and AI agent owners, in Slack): 640 chats with a gap across 18 issues, led by bulk user import. 9 tickets and 4 docs drafted.
  • Draft tickets (Linear): 9 issues drafted in Linear, from missing features to bugs.
  • Any tool you connect through MCP: Suggest an update to the AI agent’s instructions, or post the top gaps to your product channel.

Build it

Tell Rippit chat to “set up a read on what users ask the AI agents in our product to do that they can’t,” or ask your own AI assistant with Rippit connected. Rippit drafts it as a pipeline, tests it on a sample, and puts it live when you confirm. You can change what it reads, how it scores, who it notifies, and how often it runs at any time.

Questions

It reads each conversation in full: whether the agent said it couldn’t help, gave an answer that didn’t fit or failed to finish an action, and whether the user rephrased, gave up or contacted support. Each gap is checked against your docs to see if an answer already exists.
Yes. It works with an agent you built yourself, an assistant from a vendor, or agentic features that take actions for users. Decagon, Ada and Sierra connect directly, and any other agent’s conversation logs come in through the Bulk API or a CSV.
Support bot analytics track deflection and resolution. This agent looks at what users ask the AI in your product to do, and turns what it couldn’t handle into product work: missing features, bugs, knowledge gaps and improvements.
The digest goes to Slack or email. Draft issues go to Linear, or to Jira through a custom MCP connection. Draft docs go to your Intercom help center, Notion or Google Docs, all for your review. It runs weekly by default, or daily.