1 min lesson
Where non-coders actually work: the Agents window
Describe the practical point in "For knowledge work the whole demo lived in the Agents window", then say what it changes.
Step 1 of 3
When asked how you'd approach building internal GTM tooling, do not pitch a roadmap for a platform. Say: “I'd solve one painful step - consolidate the prospecting data with agent-written SQL - ship it, let reps eyeball and correct the output, then add scoring and drafting as demand pulls them in.” Incremental, demand-led, human-in-the-loop reads as someone who has actually shipped this, not designed it on a whiteboard.
Learn more
Advanced table
MCPs to your internal knowledge bases
MCPs to your internal knowledge basespersonal vs team servers
MCPModel Context Protocol. A standard that lets an AI agent pull in context from outside the repo, like Jira tickets or internal docs. Press Enter for the full definition. servers are how Cursor's team connects the agent to their internal knowledge bases. Emily splits personal MCPs (Figma, Granola) from team MCPs (Databricks for the database, Statsig to read every experiment, plus backend services). When an agent got confused mid-run, she fixed it with a single instruction: “use the Databricks MCP.” The catalog runs well past coding tools.
- MCP
- Databricks
- What it connects
- The team database - schema, user and usage data
- Scope
- Team
- MCP
- Statsig
- What it connects
- Reads through all experiments
- Scope
- Team
- MCP
- Figma
- What it connects
- Design files and design-system context
- Scope
- Personal
- MCP
- Granola
- What it connects
- Meeting notes and transcripts
- Scope
- Personal
- MCP
- Slack
- What it connects
- Fetch context and post messages
- Scope
- First-party / marketplace
- MCP
- Notion
- What it connects
- Internal knowledge base
- Scope
- First-party / marketplace
- MCP
- Datadog
- What it connects
- Alerts and monitoring
- Scope
- First-party / marketplace
| MCP | What it connects | Scope |
|---|---|---|
| Databricks | The team database - schema, user and usage data | Team |
| Statsig | Reads through all experiments | Team |
| Figma | Design files and design-system context | Personal |
| Granola | Meeting notes and transcripts | Personal |
| Slack | Fetch context and post messages | First-party / marketplace |
| Notion | Internal knowledge base | First-party / marketplace |
| Datadog | Alerts and monitoring | First-party / marketplace |
Many marketplace integrations are coding-flavored, but PM, design, writing and support ones exist too.
Explore once, then 'create a skill'pay the discovery cost once
Emily gave the agent almost no schema and let it burn tokens learning the database structure through the Databricks MCPModel Context Protocol. A standard that lets an AI agent pull in context from outside the repo, like Jira tickets or internal docs. Press Enter for the full definition.. Then: “create a skill for how to fetch user data and creation date.” Re-running, she invoked it explicitly - “use skill” - so it does not crawl the whole database again. It knows exactly where to look. Pay the discovery cost once, and every future run is fast and deterministic.
“Once it learns the structure and figures out how to work, I'll ask it to create a skill to do this. Now it runs a ton faster - instead of looking through the entire database, it uses the skill and knows exactly where to look.”
Meeting recaps and always-on automationsfrom one-off to scheduled
A directly copyable recipe: connect Granola via MCPModel Context Protocol. A standard that lets an AI agent pull in context from outside the repo, like Jira tickets or internal docs. Press Enter for the full definition. and prompt for a weekly recap of the past 7 days of meeting notes, plus an action-item list of “things I promised to do” so you stop getting pinged. The agent lists meetings, queries notes and returns a summary in seconds. It extends to Google Calendar. The output is concrete - external calls, a growth sync, a GTM-product discussion, and action items like “create a community-program application form.”
Turn that one-off into an Automation - an always-running agent on a schedule (every Monday) or a first-party trigger: a GitHub PR opened, a Slack channel created, a new message in a watched channel. Automations default to memory, so they look at prior runs, and you attach connectors inline. One live caveat: today an automation must be connected to GitHub even when the task (reading Granola notes) needs no code.
Files reports to Linear and dedupes them
Auto-upvotes when many people flag the same thing
Auto-follows-up asking for screenshots and repro steps
Runs on a schedule in a food channel
Asks what people want for dinner
Creates a DoorDash group order
Learn more
Full explanation
Full explanation
Automations currently have to be wired to GitHub even when the work is pure knowledge work with no code in sight. Know the constraint before you promise a stakeholder a code-free scheduled agent - it is a real limitation worth naming honestly.
QCursor's team built ChatGTM, an internal prospecting tool. What was their actual build approach, and why does it read as senior?
Learn more
Optional practice