Back office ops · Production

How Heidi scales globally and saves 260+ hours a month with Notion AI

The problem

As Heidi rapidly expanded globally, critical knowledge about decisions and processes lived in the heads of the people who made them, creating compounding communication overhead with every new hire, market, and product surface.

Workflow diagram · grounded in source
1
CEO yearly canons set direction
trigger
“Dr. Thomas Kelly writes what the team calls "yearly canons": strategic documents that lay out Heidi's direction and are accessible to every employee”
2
Bulletproof workspace built
integration
“Heidi built what they call their "bulletproof workspace" in Notion: connected databases rather than scattered docs, with everything searchable from a single entry point”
3
AI captures meeting decisions
ai_action
“Notion AI Meeting Notes extended this culture into conversations. Decisions, action items, and the reasoning behind them are captured automatically and made searchable for the whole company”
4
Custom Agent answers questions
ai_action
“Heidi built a Custom Agent to handle support ticket triage. It connects to their issue management software, Slack, and Gmail, and gives the broader company a fast path to answers on product plans, support themes, and what is changing. Th…”
5
Escalate uncertain queries
routing
“When the agent cannot confidently answer, the workflow escalates to the relevant project owner”
6
Knowledge gaps close over time
feedback_loop
“Each unanswered question points to a specific knowledge gap the team can close, so the system gets sharper with use”
7
MCP enables rapid prototyping
integration
“Notion's MCP integration makes the workflow practical. It connects their build environment to the tools and sources they already rely on”
Reported outcome

Notion AI saves Heidi more than 260 hours every month—200 from faster knowledge retrieval and 60+ from Custom Agent-driven workflows—while improving decision quality as the company scales.

Reported metrics
Total hours saved per month260+ hours
Hours saved from knowledge retrieval per month200 hours
hours saved from Agent-driven workflows per month60+ hours
Reported stack
Notion AINotion AI Meeting NotesCustom AgentNotion MCPClaude CodeCursorSlackGmail
Source
https://www.notion.so/customers/heidi
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Notion AI saves Heidi more than 260 hours every month—200 from faster knowledge retrieval and 60+ from Custom Agent-driven workflows—while improving decision quality as the company scales.

What tools did this team use?

Notion AI, Notion AI Meeting Notes, Custom Agent, Notion MCP, Claude Code, Cursor, Slack, Gmail.

What results were reported?

Total hours saved per month: 260+ hours; Hours saved from knowledge retrieval per month: 200 hours; hours saved from Agent-driven workflows per month: 60+ hours (source-reported, not independently verified).

How is this back office ops AI workflow structured?

CEO yearly canons set direction → Bulletproof workspace built → AI captures meeting decisions → Custom Agent answers questions → Escalate uncertain queries → Knowledge gaps close over time → MCP enables rapid prototyping.

Related back office ops cases