Elysium
The Threshold Collective

A wellbeing companion built on lived experience, not surveillance.

Elysium is the intelligence layer of The Threshold Collective. It validates before it advises, refuses employee monitoring on principle, and routes crisis language to real human services. This is a public V1 demo. Responses are paced because the model is running on modest infrastructure, and because considered answers take time.

Model Elysium V1
Guardrails Crisis routing, PII scrub
Governance Lived experience led
Status Public preview
Loading…
Elysium
Elysium is temporarily unavailable while we complete a planned infrastructure migration.
If you need support right now, please speak to a real person.
Samaritans 116 123  ·  Shout text SHOUT to 85258
NHS urgent mental health line 111, option 2  ·  Emergency 999
Non-negotiables

Five principles Elysium will not compromise

01

Ethics first

Every design decision runs through a governance check before utility. Speed is not an excuse.

02

Privacy first

Elysium will not profile employees, flag sentiment, or build shadow wellbeing scores for managers.

03

Lived experience governed

The people who built this have been on psychiatric wards. That shapes what the model refuses to do.

04

Transparent by default

When a guardrail fires, you see it. When the model is uncertain, it says so. No hidden logic.

05

Harm reduction before convenience

Crisis language is escalated to human services. Elysium will not role-play as a therapist.