Value chain
R&D / Engineering
R&D builds the product. Its output quality and speed affect every downstream department, from what Marketing can promise to how many tickets Support gets.
Key processes
- Sprint planning
- Development and code review
- Testing and QA
- Release management
- Technical debt management
KPIs that matter
- Deployment frequency
- How often the team ships to production.
- Change failure rate
- Share of releases that cause incidents.
What AI agents can do here
- Code completion and generation
- Writing tests
- Explaining legacy code
- Agents fixing bugs end to end from a ticket
- Automated code review
- Agent teams building whole features with humans setting direction and reviewing
What good looks like
- Small, frequent releases behind feature flags, with automated tests on every change.
- Measure deployment frequency, lead time, change failure rate and time to restore together.
- Reserve a fixed share of capacity for maintenance and technical debt.
Common pitfalls
- Optimizing velocity while quality silently erodes
- Big-bang releases that are hard to roll back
- Unreviewed changes to production under deadline pressure
Agent tasks and the human control they need
| Task | Maturity | Human control |
|---|---|---|
| Code completion, test generation and code explanation | Today | Every change is reviewed by a person before merge. |
| Fix well-specified bugs end to end from a ticket | Emerging | Changes go through the same review and CI as human code. |
| Agent teams building whole features from a product brief | Future (3–5 years) | Humans set direction, review architecture and own releases. |
Today: in common production use. Emerging: piloted by early adopters. Future: plausible within 3–5 years, not yet dependable.
End-to-end processes this department is part of
Related processes
Incident management
Detect → Triage severity → Contain → Resolve → Communicate → Blameless post-mortem
Mature teams measure mean time to recovery (MTTR), not just how rarely things fail.
Idea-to-market
Discovery → Prioritization → Design → Build → Test → Launch → Measure adoption
Strong product teams kill more ideas than they ship, saying no protects R&D capacity.
Inputs: Roadmap from Product, Infrastructure from Run Ops. Outputs: Releases and product quality for Marketing and Sales.