CASE 03 · 2025 · Co-builder · Multi-agent AI
SENTINEL
Privacy-first enterprise wellbeing analytics with a 3-agent AI orchestra.
STATUS
PRIVATE WORKSPACE — CO-BUILT
RESULT
Three specialist agents answering plain-language workforce questions in ~300ms — with privacy intact by architecture.
LIMITATIONS
Evaluated on a synthetic corpus matching production shape; no live enterprise deployment shown.
01
THE PROBLEM
Burnout is a business risk HR can't see until it's expensive. Any solution had to be privacy-first — anonymized signals, not surveillance — and answer plain-language questions, not just dashboards.
3 specialized AI agents in one system
02
THE HARD PART
Keeping natural-language queries over workforce analytics under a second — pre-aggregating graph features per team turned seconds of latency into ~300ms without losing nuance.
03
WHAT SHIPPED
Three specialized agents (burnout, talent, team health) on anonymized signals only — plain-language answers over workforce analytics, privacy intact by architecture.
THE FULL STORY
The story
Burnout is the most expensive invisible problem in a company — and the moment you
build anything to measure it, you're one bad decision away from surveillance. So the
privacy constraint was set before a single line of code: anonymized patterns only,
no identity in the pipeline, by architecture not policy.
Three different questions needed three different data shapes. Burnout scoring wants
individual behavioural signals; talent discovery wants strengths and patterns; team
health wants collaboration structure. One model would blur all three. Three
specialists with a routing layer keep each one small, testable, and accountable.
The latency fight was the engineering story. Natural-language queries over
workforce analytics started at seconds. Pre-aggregating graph features per team
nightly turned that into ~300ms — the same insight, the same nuance, just arranged
before anyone asks.
IMPACT
- 3 specialized AI agents in one system
- Graph analysis for team-collaboration insight
- Privacy-first: anonymized patterns only
- Natural-language queries over analytics
THE ARCHITECTURE
Layered, labelled, honest — the system as it actually stands.
01 · SIGNAL INTAKE
02 · GRAPH LAYER
03 · AGENT ORCHESTRA
04 · INTERFACE
SYSTEM FLOW
WHAT IF …
Ask the project a different question. The architecture has to defend itself.
What if you used a single LLM instead of 3 agents?
One model doing scoring + talent + health would blur the evidence trails and make audits impossible — and hallucinate team conclusions. The three-agent split keeps each system small, testable, and accountable; the orchestrator only routes. More moving parts, but each one is honest.
What if the board demanded per-person scores?
Refuse, architecturally. The whole design is privacy-first: anonymized patterns only, no identity in the pipeline. I'd show them cohort-level risk trends, team-collaboration graphs, and department aggregates — the same insight without the surveillance.
DIVE DEEPER
Built it — now the descent. Each question opens the next layer: why, why this architecture, what broke, what I'd change.
01Why three agents?
Burnout risk, talent discovery and team health are three different questions with different data shapes. One model would blur them; three specialists stay honest.
02Why privacy-first from day one?
Wellbeing analytics dies the moment it feels like surveillance. Anonymized patterns only — no names attached to scores.
03What went wrong?
Natural-language queries were slow until I pre-aggregated graph features per team. Latency dropped from seconds to ~300ms.
04What would I do now?
Add temporal decay to signals and a human-in-the-loop review queue before any score leaves the system.
THE REPO, INSIDE
Not a screenshot — a live iframe pulling this repo's README straight from the CDN.
THIS PROJECT LIVES IN A PRIVATE WORKSPACE