Scaling Expertise in a Multi-Stakeholder Care Network

The Challenge

Supporting families with neurodiverse members requires navigating a complex ecosystem: coaches, cantonal funding bodies, health insurers, schools, therapists, and other specialists. Each stakeholder holds a piece of the puzzle — but the knowledge was fragmented, compartmentalized, and hard to access when families needed it most.

Coaches, mostly working 80% with 2-3 appointments daily, spent too much time on administration and coordination. Families faced long waiting times before receiving support. And the rich experiential knowledge that coaches accumulated over years? It lived in their heads, their notes, their supervision conversations — rarely captured in a form that could help the next coach facing a similar case.

The Knowledge Landscape

This was predominantly experiential territory:

• Structure: Mostly unstructured — reports, session notes, voice transcripts. Some semi-structured objectives and progress measures. Very little truly structured data beyond hours and budgets.
• Provenance: Internal methods and budgets, shared-private progress updates for cantons and insurers, and external knowledge bases for methodologies and clinical frameworks.
• Knowledge Types: Heavily experiential — what works for which family, in which context, over what timeframe. Some stochastic patterns around intervention effectiveness. Minimal deterministic elements beyond approvals and financials.

The Approach

We started by structuring knowledge along capabilities and expertise levels — not to replace judgment, but to make it accessible.

Rollout was deliberately phased:

1. Internal champions and management (who were also experienced coaches)
2. Friends-and-family clients willing to pilot
3. Broader rollout to families in scope

The mantra: “Better something now that gets you 60% of the way than waiting six months for perfect.”

From day one, we implemented:

• AI-assisted intake interviews with 1-3 quality gates depending on case complexity — reducing time-to-first-contact while ensuring human coaches focused where they added most value
• 8 guardrail categories covering sensitive indicators from self-harm to domestic violence, each mapped to a 9-box escalation matrix using keyword detection, pattern recognition, and temporal graphs across sessions
• Human-in-the-loop review calibrated by severity: immediate for critical flags, next-day for moderate, or at next case touchpoint for lower-priority signals
• Supervision integration: AI now prepares supervision sessions with case summaries and pattern observations; supervision dialogue feeds back into the system, continuously extending the knowledge base

The Impact

Metric	Result
Family waiting times	–60%
Upfront time/budget commitment	–50%
Progress tracking	From 3rd session (once goals set)
Time to internal testing	3 months
Time to measurable impact	Immediate after rollout

The Human Story

The team was split. Most coaches were skeptical — understandably protective of their craft and their relationships with families. But 2-3 champions, including experienced coach-managers, saw a different possibility: less time on admin, more time face-to-face with the families who needed them.

They won over the others gradually. Not with promises, but with results. And one commitment that mattered: no one gets replaced. Demand for these services far exceeds capacity. The consequence of this work isn’t efficiency for its own sake — it’s growth. More families reached. More coaches supported. More expertise shared.