Create an agent
Open the AI Chatbots module to see your existing agents and their deployment status at a glance — no code required. Choose Create Agent, give it a name, and pick the knowledge it should learn from.
- Knowledge agents answer questions from DataHub documents — PDF, DOCX, TXT, and MD files
- Analysis agents answer questions from a connected analysis dataset — CSV/XLSX files or CRM and Sheets sources
- Website ingestion lets an agent crawl same-domain pages and index the text so it can answer website-specific questions
See AI Chatbots in action
Watch: create an agent, train it on your sources, and get source-grounded answers.
Train it on your sources
Because each agent is grounded in the sources you connect, answers come from your own content — not generic web results — and include citations. When you update a document or re-sync a source, the agent's knowledge reflects it.
Deploy it three ways
Embed it on your website with one line of code, deploy it to Slack so your team can chat with it, or publish it internally to share it across teams inside Intelligence Navigator. Track conversations and activity to see what people ask and refine the agent over time.
Best practices
- Keep each agent focused on a clear topic — support, onboarding, internal knowledge — for the most accurate answers
- Connect only the sources relevant to that topic so responses stay grounded and on-brand
- Review conversation transcripts regularly and add sources where the agent struggled
Next steps
Once your agents are live, the AI Operations Center helps you monitor and coordinate them in one place.