AI transformation of product operations
Agentic workflows for research, feedback triage, documentation and specs, run by the whole product team.
What I've done
- Moved product operations onto agentic workflows: research and data analysis, customer feedback triage, documentation updates and formatting, specification to tickets
- Wrote each one as a reusable specification the whole product team runs, rather than a personal prompt
- Put a measured accuracy loop behind the workflow that touches customer-facing documentation, so its quality is a number rather than a claim
- Reinvested the time it returns into discovery
- Ship production code alongside the engineers, from interface fixes to architecture changes
Rough size
- Agents in daily use16
- Recurring workflows replaced6
- Product team running them16 people
- Company size110 people
Impact metrics
- 10 hrs
- Returned per week, per product manager
- 92%
- Documentation accuracy, up from 80%
- 42
- Pull requests merged to production
The time returned goes back into customer research, which is the point of doing it.