AI · Agile
Co-built an AI retrospective facilitator
An AI co-facilitator for team retrospectives
Co-developed YodaAI, an AI-powered retrospective co-facilitator grounded in PMI Disciplined Agile and positioned as a helper that earns trust instead of an automation that replaces facilitators.
YodaAI adds range to the portfolio by showing product strategy in AI and facilitation tooling, not just infrastructure-heavy domains.
Challenge
Retrospectives are one of the most valuable agile rituals and one of the most inconsistently facilitated. Teams struggle with shallow insights, repeated patterns, and action items that disappear after the meeting.
Context
Scrum Masters, project managers, and team leads who facilitate retrospectives and want deeper, more actionable insight without adding facilitation overhead.
Operating constraints
->The product needed to align with PMI Disciplined Agile instead of generic Scrum to create real differentiation.
->AI had to feel like a trusted co-facilitator rather than surveillance in a psychologically sensitive space.
->Co-development with a technical partner required clean decision boundaries to avoid drift and paralysis.
->The concept needed market validation before backend infrastructure investment.
Strategic approach
Anchored the product in PMI Disciplined Agile
That gave the facilitation model a structured knowledge base and a clearer market position than a generic AI retro assistant would have had.
Designed the AI as a co-facilitator, not an autopilot
The AI suggests prompts, themes, and patterns, but a human facilitator controls the flow. That protects psychological safety while still increasing insight density.
Validated through presentation-first testing
A 22-slide deck with live Mentimeter polling let the team test resonance with practicing PMs before committing to deeper engineering investment.
My role
Co-developed the product concept with Richard Magu. Defined the product vision, UX model, and DA integration, built the presentation narrative, and managed beta-testing priorities after the PMI Kenya presentation.
Results and impact
- *Presented the concept to the PMI Kenya Chapter and generated clear interest from practicing PMs and Scrum Masters.
- *Used live Mentimeter polling to validate that pattern blindness and action-item follow-through were real audience pain points.
- *Moved into beta testing with feedback shaping the next iteration of the facilitation engine.
- *Established a sustainable co-development model with clear role boundaries.
Reflection
Two things stood out: partnership needs explicit decision boundaries, and AI products earn trust in layers. Starting as a helpful copilot was stronger than trying to arrive as a fully autonomous facilitator.