The Kepler Assistant

Chat with your CRM — ask questions across your whole workspace, @-mention records, and share conversations with the team.

The Assistant is a chat interface over your whole workspace. Ask it questions the way you'd ask a colleague who somehow remembers every record — "who have we placed at fintech companies in Leeds?", "show me contract DevOps candidates available this month", "what's the state of the Acme account?"

Starting a chat

Open Assistant in the sidebar and type into the composer, or select Ask AI — or press Cmd/Ctrl + Enter — in global search to open the Assistant, then enter your question there. The new-chat screen offers your most recent conversation as a jumping-off point. Answers stream in as formatted text, above a single activity line saying what the assistant is doing as it works ("Searching candidates", "Loading record"…) — expand it to see each step.

What it can do

  • Search candidates semantically — over profiles, CV content and recruiter notes, with structured filters (status, availability, location, source, dates) layered on top.
  • Search all records — jobs, companies, contacts, applications, placements, tasks and lists.
  • Count and list — "how many candidates did we add in June?"
  • Drill into a record — full detail on a candidate (profile, recent notes, pipeline) or any job, company or contact.

The Assistant can read and search CRM data, and can propose changes — including notes, tasks, stage moves, record updates and new records. Nothing in a proposal changes the CRM until you confirm it.

@-mention records for precision

Type @ in the composer to search and attach specific records to your message. Mentioned records appear as chips, and the assistant works from the exact records you pointed at instead of guessing by name — essential when you have three John Smiths.

Chat history and sharing

  • The history popover lists My chats and Shared with team.
  • Each conversation can be renamed, deleted, and toggled between private and shared with team — share a research thread so a colleague can read your working.
  • Only the chat's owner can send messages in it; teammates viewing a shared chat read along.

Getting better answers

  • Say what you actually want to do next — "…so I can build a call list" gets more useful output than a bare query.
  • Use @-mentions whenever the subject is a specific record.
  • Use Copy on a finished answer to lift a shortlist or summary straight into an email.

The assistant sees only what you can see — it respects your workspace membership and email sharing settings. See AI & data privacy for how conversation content is handled.

Read as Markdown