Building an AI-powered skills platform from zero to global adoption
Took an enterprise skills and customer-delivery platform from concept to global launch — improving how Microsoft matched consultants to customer needs while creating the foundation for AI-assisted skill discovery.
improvement in consultant-customer match efficiency
improvement in skill discoverability
monthly active user adoption
NPS after global launch
Context
Microsoft's customer-delivery organization had deep expertise, but finding the right consultant for a customer need was slowed by fragmented skill data and manual coordination. Better search alone would not solve it: the product had to make skills easier to express, trust and act on.
I led the product from zero to one across business, design, engineering and operations — defining the roadmap and MVP, building the customer-research loop, and taking the platform through a five-ring global rollout.
Visual model
Data Intelligence Platform
Self-serve agent creation, grounding and evaluation
- Copilot agent experiences
- Contextual grounding
- Personalization
- Enterprise data workflows
15+ Copilot teams building on one capability · 600M+ MAU served
A shared skills foundation connecting consultant signals, customer needs and delivery decisions.
The product bet
The real problem was not a missing directory. It was a matching problem: customer needs were expressed one way, consultant expertise another, and the organization paid the translation cost every time.
I framed the product around a measurable decision — helping delivery teams identify the right expertise faster — rather than around collecting more profile data. That gave the team a clear outcome for the MVP and a foundation for predictive and Copilot-enabled discovery.
How I took it from idea to adoption
- 01
Aligned more than ten partner teams on the problem, MVP and long-term product vision.
- 02
Established a customer-research network and feedback loops to validate hypotheses before and after launch.
- 03
Built a predictive analytics workspace so delivery leaders could move from reporting to decisions.
- 04
Conceptualized an early Copilot-enabled skill-search experience, using natural interaction to reduce discovery friction.
- 05
Managed a five-ring global rollout, using adoption and customer satisfaction — not launch completion — as the success bar.
Why this story matters
This was the product experience that taught me how AI earns a place inside an enterprise workflow. The intelligence matters, but only after the data, incentives, operating model and user trust are strong enough for people to act on its recommendation.
The result combined business impact and product health: faster matching, better discovery, 85% monthly active adoption and a 91 NPS.
Themes
- 0→1 product
- Enterprise AI
- Marketplace matching
- Customer discovery
- Predictive analytics
- Global launch
- Adoption
- Cross-functional leadership
The strongest enterprise AI products do not begin with a model. They begin with a decision people need to make better.