Long profile
A comprehensive, evolving record: medical restrictions, goals, assessments, posture, feedback and attendance, generated by LLM summarisation and refreshed nightly or on new data.
- Home
- Case Studies
- FitwithPari
Case study · Fitness / Coaching / HealthTech
Building the intelligence layer for personalised coaching at scale.
CoachPrep AI Agent (Phase 1) turns raw student data into pre-computed profiles, so coaches get instant, class-ready insight with zero AI latency in session.
- Client
- FitwithPari
- Industry
- Fitness / Coaching / HealthTech
- Delivery
- AI profile intelligence platform
- Capability
- AI Agents & Automation
Results at a glance
0 msreal-time AI latency during classes
The challenge
What FitwithPari was up against.
Coaches collected onboarding forms, assessments, posture notes, weekly feedback and attendance, but in raw, fragmented formats that were hard to use before or during class.
- Understand a student quickly before class
- Personalise sessions in real time
- Track progress across weeks
- Scale quality coaching across many students
The solution
What we built.
Skugal designed an agent-ready foundation that builds high-quality profiles in advance and separates deep understanding from quick, class-ready insight.
Short profile
Class-specific must-know, should-know and nice-to-know summaries for each format (strength, yoga, mobility), pre-computed before class.
Incremental updates
New feedback refreshes the long profile and regenerates only the affected short profiles.
Supabase + n8n foundation
Versioned profiles with audit trails and deterministic, retry-safe workflows that future agents can reuse.
How it works
From input to outcome.
- Raw student data (Supabase)
- Full profile builder
- Short profile generator
- Incremental updates
- Instant coach view
Business impact
What changed.
By the end of Phase 1, FitwithPari had a stable AI profile engine and a scalable foundation for agent-driven coaching.
Technology
- LLM summarisation
- Supabase (Postgres)
- n8n orchestration
- RPC atomic upserts
- Instant coach access to student insight
- Zero real-time AI latency during classes
- Consistent understanding across coaches
- Profiles that evolve automatically with progress
CoachPrep is no longer an experiment. It’s the intelligence layer behind the coaching.
Next step
Have a similar challenge?
Tell us about the workflow or product. We’ll come back with an architecture, a delivery plan and where AI agents fit.