Technical evidence
How Bondlyy works, end to end
A live product, not a concept. This page documents the system architecture, the matching logic and the development roadmap.
The Bondlyy System
Five layers. Each one is running in the product today.
Intake
Every client completes a wellbeing profile: interests, personality, conversation style, preferred experiences, language, location, mobility, confidence and desired frequency.
Match Engine
Eight weighted dimensions produce a 0-100 Bondlyy Compatibility Score with a readable per-dimension breakdown. Never a black box.
Structured sessions
Once a companion is assigned, a session plan is generated from the client's profile: goals, designed activities, conversation starters and pacing notes.
Outcome measurement
After each session the client rates mood before/after, felt connection and satisfaction. These roll up into the Bondlyy Connection Index™, month by month.
Learning loop
Outcome scores are grouped by match band. Stronger matches producing stronger outcomes validates the model and feeds reweighting of the engine.
The matching logic
A client's wellbeing profile is scored against every active companion across eight weighted dimensions, producing a 0-100 Bondlyy Compatibility Score. Users see exactly how their score was built: every dimension shows its own score and a plain-English note.
- Interests match through a keyword model, robust to wording differences
- Personality blends trait overlap (60%) with conversation-rhythm affinity (40%)
- Language, location and availability are hard practical filters, weighted in
- Confidence and frequency dimensions protect fragile first sessions
- Bands: 80+ Excellent, 65+ Strong, 50+ Good, below 50 Worth exploring
Dimension weights
Architecture
Four layers, secured end to end. Client apps talk to a role-aware platform layer; services are isolated behind it; all data sits under row-level security.
Client apps
Platform layer
Service layer
Data layer
Learning loop: Connection Index outcomes flow back into the Match Engine, closing the cycle between layers 4 and 2.
Development roadmap
Staged from live pilot to national scale.
Pilot. Live now.
- Matching, booking, payments, messaging and outcome measurement all running in production
- Companion vetting pipeline: application, ID check, DBS, references, video interview, training
- Connection Index™ live per client; admin outcome analytics by match band
- Organisation dashboards for referral partners with anonymised outcomes
Prove the loop
- Outcome data by match band (already collecting) used to reweight the Match Engine
- Automated matching suggestions for admins, ranked by compatibility score
- Companion quality scores from paired outcome history
- Monthly rollups surfaced to clients in their own progress charts
City rollout
- Expand companion supply city by city with local vetting pipelines
- Institutional partnerships: social prescribing, care homes, councils
- Subscription tiers for recurring companionship
- Safeguarding tooling: incident register, escalation SLAs, referral routes
National scale
- The Learning Loop as the product: outcome-verified matching no competitor can copy
- Outcome reporting API for referring organisations and funders
- Coverage targets by region with local Connection Index benchmarks
