AI in Business
Lessons from integrating AI into Atlassian teams, startups like Weel, and my own products.1. Identify High-Leverage Workflows
- Repetitive decision support (support triage, finance approvals).
- Knowledge-heavy tasks (docs search, preventative playbooks).
- Creativity boosters (campaign ideation, storytelling outlines).
2. Build a Thin Slice
- Choose one workflow, instrument baseline metrics.
- Ship an internal copilot/agent to a small cohort.
- Compare outcomes vs control: time saved, quality, satisfaction.
3. Governance
- Define data boundaries, retention policy, review process.
- Add audit logs + human approval for risky actions (finance, customer comms).
- Track cost per request + ROI.
4. Change Management
- Train teams (Loom library + office hours).
- Celebrate wins with short stories (Agent Diaries format).
- Collect friction logs; iterate weekly.
5. Case Notes
- Weel: AI-assisted approvals cut review time by 45%.
- Bugcrowd: Triage copilot summarizing reports before human review.
- MetaLabs: Launching AI support for builder community; adoption tracked via PostHog funnels.