Skip to main content
You ask in plain language; your agent uses the Jedify tools to answer from your data. Five examples:

​
See what you can ask

Prompt: “List 5 questions I can answer about our data with Jedify.” Agent: calls contextualize and get_semantic_layer_entities to read your semantic layer. Result: concrete, answerable questions grounded in your data — not generic suggestions.

​
Ask a direct question

Prompt: “What was total invoiced revenue in USD for the first half of 2026?” Agent: calls ask_a_single_question — Jedify generates the query, runs it on your warehouse, and returns the data. Result: the figure you asked for.

​
Ask one of your custom agents

Prompt: “Ask our revenue agent why Q2 bookings came in below plan.” Agent: calls list_agents to find the agent, then ask_an_agent — the agent answers with its own persona, instructions, and entity scope applied. Result: an answer shaped by the expert your team built, not a generic query.

​
Combine your data with the web

Prompt: “Pull our invoiced revenue by region for 2026, compare it to public industry growth rates, and write a short brief.” Agent: uses ask_questions for the internal numbers and its own web search for the benchmarks. Result: a brief combining your figures with outside context. For deeper internal investigations (“why did X change?”), the agent uses ask_a_research_question.

​
Share an answer

Prompt: “Send me a link to that answer so I can forward it to finance.” Agent: waits for the session’s questions to finish, then calls create_share_link. Result: a link that opens the full answer — narrative, charts, and data — in the Jedify app. Recipients need a Jedify login in the same account.

​
Tips

  • Be specific with dates, metric names, and segments.
  • Start with the headline number, then ask the agent to break it down.
  • Say “research” or “investigate” to trigger a multi-step analysis.
  • If your account has more than one data source, name the source you mean — your agent passes it as data_source_id.