Case studies / Felix
Felix
- Discipline
- Product strategy, AI and design
- My role
- UX and Design Strategy Owner
- Worked with
- Executive, product and engineering leaders
- Stage
- Early stage development
Confidential: names are anonymised and some details simplified.
01 Summary
The vision for an AI first platform.
As UX and Design Strategy Owner, I set the product vision for Felix, an agentic AI platform, and prototyped it in Cursor and Claude. The concepts were adopted into future product direction.
15 to 20 min
to investigate a single issue, before anyone could act.
02 The problem
Five systems, one overloaded team.
Staffing, compliance, fulfilment, forecasting and payroll all ran on separate systems. Investigating one issue took 15 to 20 minutes before anyone could act. The only way to scale was more headcount.
03 The process
From chatbot to operating system.
Four steps, each one a little closer to a system that runs the operation with people in control.
01
Ask
Felix started life as a Q&A chatbot. Useful, but people still had to know what to ask. I reframed it into something that acts, not just answers.

01 · Ask · where Felix started 02
Agents
Behind one front door, specialist agents each own an area: staffing, compliance, fulfilment, forecasting, payroll and performance. People ask once, and the right agent picks it up.

02 · Agents · one front door 03
System
I set four rules for trustworthy AI: decision engine, transparent automation, human control and one ecosystem. Those rules pulled six connected experiences into one system.

03 · System · low fidelity concept 04
Prototype
I built it in Cursor and Claude, then took executive, product and engineering leaders through real workflows. Everyone ended up backing one operating model.

04 · Prototype · first working version
04 The metrics
15 to 20 min
to investigate one issue before Felix
6
connected experiences in one platform
1
shared operating model, backed by leadership
05 The solution
Problems spotted, reasoning shown, people in charge.
An AI first operating system that spots problems, explains its reasoning and recommends the fix, with people approving the big calls.

It set the organisation's AI operations vision and became the basis for roadmap planning.

Next case study
Indeed / Indeed Flex