04

AI-assisted knowledge operations

Project Mosaic

A working prototype for creating a searchable, durable record of shipped product work at Wesco International through structured contributions, AI-assisted organization, and on-demand discovery.

RoleProduct lead & system designer
ScopeSeven product scrum teams
StatusWorking prototype
PlatformSharePoint · Microsoft Copilot

The problem

Three hours of demos still left people searching for answers.

Each month, seven product scrum teams presented their recently delivered work to the business. Product owners and managers explained the releases, and developers frequently demonstrated the features. The meeting lasted roughly three hours and pulled a large cross-functional audience into the same live session.

The recordings preserved the meeting, but not the knowledge in a useful form. They were difficult to search, feature details were buried in hours of video, and attendees often asked questions afterward that the demonstrations had already answered. The process was expensive, time-consuming, and too dependent on people being present at the right moment.

Conceptual information flow

A portfolio-safe representation of the system—not a reproduction of proprietary Wesco screens or content.

01

Submit

Product owners use a SharePoint form to capture the feature, team, release, purpose, documentation, and video evidence.

02

Interpret

Microsoft Copilot agents and workflows condense, classify, and normalize the submitted release information.

03

Organize

Structured records create a durable product history organized by team, release, feature, and business purpose.

04

Discover

A searchable dashboard lets employees find what shipped, when it shipped, why it mattered, and where to learn more.

What I designed

01

Structured contribution

Replaced free-form live presentation preparation with a consistent intake model that captured the context people needed later.

02

AI-assisted curation

Used Copilot agents, workflows, and decision logic to turn mixed documentation and video submissions into clear, organized release records.

03

Durable product memory

Designed the repository to chronicle the platform as it evolved instead of allowing valuable release knowledge to disappear inside meeting recordings.

04

Self-service visibility

Made feature discovery available on demand so leaders, business partners, and delivery teams could answer questions without replaying a meeting or locating a presenter.

The prototype result

A working prototype showed how demonstrations could become reusable organizational knowledge.

The prototype demonstrated a path away from a single synchronous event and toward a living internal catalog. Each contribution could serve multiple audiences and remain useful long after the original release.

The system was functional and received limited internal use, but it did not progress to broad organizational deployment before I left Wesco. I therefore describe it as a working prototype rather than a completed transformation.

The opportunity it demonstrated was leverage: reduce recurring meeting cost, preserve feature history, improve transparency, and let employees discover product information when their work required it.

No proprietary screenshots or internal Wesco information are reproduced in this case study. The visual model is conceptual.

Design principles

01

Asynchronous by default

Reserve live time for decisions and discussion, not information that can be contributed and consumed independently.

02

Retain the evidence

Keep documentation and demonstrations connected to the summarized record so users can move from overview to source material.

03

Design for retrieval

Capture metadata at submission so search and filtering are part of the operating model rather than an afterthought.

04

Use AI with structure

Give automation clear inputs, decisions, and destinations so it improves consistency without obscuring ownership.

Return to

Selected work