Blog

1 min readtutorial

How to put AI agents into tools your team already uses

A practical approach to embedding agents in Gmail, CRM, docs, and chat - instead of forcing a new platform day one.

Written by Dali

Dali is an AI agent systems studio. David leads engineering and product systems; Liana leads operations and workflow fit. We ship production agents inside tools teams already use.

David Hakobyan · Dali

How to put AI agents into tools your team already uses

Direct answer

Put agents where the work already happens. If intake lives in email and CRM, the agent should read and write there - not force every user into a new portal. New platforms can come later; first win is removing hops.

Default surfaces

  • Inbox and calendar for communication work
  • CRM for pipeline state
  • Docs / wiki for knowledge
  • Chat for internal ops signals
  • Sheets for lightweight queues when that is the real system of record

Integration pattern

  1. Observe: what tool is source of truth for each field?
  2. Act: which writes are drafts vs live?
  3. Gate: which writes need human approval?
  4. Notify: where does the human see the queue?
  5. Log: where do we store agent decisions?

Anti-patterns

  • Building a parallel "AI workspace" nobody opens
  • Syncing every field both ways on day one
  • Letting the agent invent CRM stages not in your process

Dali's bias

We implement production agents inside tools teams already run. See the solutions map for multi-agent production patterns.

FAQ

  • Usually no. Start with the tools of record and a thin orchestration layer.