Update from code changes: Deep Research mode in AI Agent composer

[mintlify]

Summary

Adds a new AI agent guide covering Deep research — the in-product counterpart to the Deep Research API — now available as an inline composer mode. The guide covers availability, the composer control, preflight and depth tiers, the durable run card, the linear markdown report with inline charts (snapshotted with a live-refresh option), and follow-up actions.

Changes

• New guide guides/ai-agents/deep-research.mdx covering availability (ai-deep-research feature flag, Enterprise-only), the composer Ask / Deep research switch, the preflight depth picker (Low / Medium / High / Extra High → low/medium/high/xhigh), evidence sources, the durable run card lifecycle, the linear markdown report with confidence-tagged findings and inline snapshot charts (with per-chart View live data refresh and an Agent-computed badge for derived charts), and follow-up actions.
• Added the new page to the AI agents nav group in docs.json, just after Using AI agents.

Context

• Upstream source PRs:
  • feat(ai): add Deep Research mode to AI Agent lightdash#25613 — "feat(ai): add Deep Research mode to AI Agent"
  • feat(ai): produce deep research reports as markdown lightdash#25732 — "feat(ai): produce deep research reports as markdown"
  • feat(ai): render deep research reports as one markdown flow lightdash#25733 — "feat(ai): render deep research reports as one markdown flow"
  • feat(ai): link deep research runs to agent threads lightdash#25734 — "feat(ai): link deep research runs to agent threads"
  • feat(ai): snapshot deep research charts with on-demand live refresh lightdash#25754 — "feat(ai): snapshot deep research charts with on-demand live refresh"

Update history

• Initial commit: added the Deep research guide covering the composer control, preflight tiers, run card lifecycle, structured report view, and follow-up actions. Source: feat(ai): add Deep Research mode to AI Agent lightdash#25613.
• Follow-up: rewrote the report section to reflect the shift to a single markdown flow with inline callouts and confidence-tagged findings; documented snapshotted charts, the per-chart View live data refresh, the Agent-computed badge for inline charts, and server-side thread linkage so runs are visible on any browser or device. Sources: feat(ai): additive viz renderer and query-results plumbing for deep research lightdash#25730, feat(ai): produce deep research reports as markdown lightdash#25732, feat(ai): render deep research reports as one markdown flow lightdash#25733, feat(ai): link deep research runs to agent threads lightdash#25734, feat(ai): snapshot deep research charts with on-demand live refresh lightdash#25754.

[github-actions]

🤖 Documentation Bot

✅ All checks passed! No issues found.

[mintlify]

Preview deployment for your docs. Learn more about Mintlify Previews.

Project	Status	Preview	Updated (UTC)
lightdash	🟢 Ready	View Preview	Jul 17, 2026, 12:45 AM

[mintlify]

Update from code changes updated this PR.

Rewrote the Deep research report section to cover the markdown flow, inline snapshot charts, live refresh, and thread-linked persistence.

Source PRs

• lightdash/lightdash#25613 — feat(ai): add Deep Research mode to AI Agent
• lightdash/lightdash#25730 — feat(ai): additive viz renderer and query-results plumbing for deep research
• lightdash/lightdash#25732 — feat(ai): produce deep research reports as markdown
• lightdash/lightdash#25733 — feat(ai): render deep research reports as one markdown flow
• lightdash/lightdash#25734 — feat(ai): link deep research runs to agent threads
• lightdash/lightdash#25754 — feat(ai): snapshot deep research charts with on-demand live refresh

[github-actions]

🤖 Documentation Bot

✅ All checks passed! No issues found.