Notion AI Agent — 7 Powerful Things It Can Do for You

About this video

Garrick, in a video made in partnership with Notion, introduces Notion’s new AI Agent as a step beyond the older Notion AI: instead of only summarizing or suggesting edits, the agent takes a goal, plans steps, and executes them inside the workspace — creating and editing pages and databases, searching across pages, updating multiple documents, and following user instructions and style preferences over time. Note: the title’s ‘7 powerful things’ framing is not enumerated as seven distinct items in the transcript; the video instead walks through worked examples. First, messy meeting notes are summarized into key decisions, blockers and next steps, and the agent creates a new weekly-update page itself rather than leaving copy-paste to the user. Second, in a project tracker database, the agent reviews due dates, creates a new ‘overdue’ status value that did not exist, marks past-due tasks, then analyzes owners’ high-priority loads — it identifies three owners as overloaded, suggests rebalancing options, and offers to reassign tasks and add a grouped/filtered view. Third, asked to launch a two-week social media campaign for an app update, the agent builds from scratch a campaign calendar database (posting dates, platforms, post titles) plus a separate page of editable draft posts mixing announcements, tips and engagement content, then turns its suggested next steps into a checklist page. The agent panel shows its thought process, lets users pick the model (Claude or ChatGPT at the time, with more promised) and control sources such as web search. Garrick cautions users should vet and customize everything the agent generates.

Key takeaways

• The new Notion AI Agent executes multi-step tasks end-to-end (creating pages/databases and updating records), unlike earlier Notion AI which only summarized or suggested.
• It can turn messy meeting notes into structured decisions/blockers/next steps and automatically create the shareable weekly-update page.
• On databases it can invent missing status values, mark overdue tasks at scale, analyze workload balance, and offer to reassign tasks and build monitoring views.
• It can scaffold a whole project from a prompt alone — e.g. a two-week social campaign calendar database, draft posts page, and a next-steps checklist.
• Users can choose the underlying model (Claude or ChatGPT), restrict information sources, watch the agent’s reasoning, and should still vet/customize its output.

Episode details: published October 21, 2025 · 11:50 long. Summary written by Self Growth Videos from the episode itself. Business results and any income figures mentioned are Kevin Stratvert’s own claims, shared for education — they are not typical or guaranteed.

← Previous: The Ugly Truth About AI Agents (and How to Spot the Real Ones) · Next: Jotform Instagram Agent Tutorial – How to Automate Instagram DMs with AI →

All Kevin Stratvert videos · Kevin Stratvert profile · Kevin Stratvert on YouTube