Safeguarding
What happens when a conversation gets serious?
The coach recognises it, changes mode, and signposts real support. That behaviour is built in, tested, and measured.
Try it yourself. Open a session and start with “I'm feeling really down…” — then watch how the coach responds. The heavier the conversation becomes, the more carefully it's handled.
Built in, not bolted on
Safeguarding runs through three layers of the product. Remove any one of them and the other two still hold.
The coach's core instructions
Safeguarding behaviour is written into the foundation of every coach — the instructions that shape every response, in every session, in every language. It cannot be switched off by anything a user says.
Crisis modes
If a conversation signals that someone needs more than coaching, the session changes. The coach stops coaching, acknowledges what's happening, and signposts appropriate support. These modes are tested, not theoretical.
Measured every session
Our quality framework, My Conversation Blueprint, includes dedicated safeguarding parameters. Sessions aren't just delivered safely — they're scored against safety, continuously.
Coaching, to a professional standard
Our coaches are built to adhere to the ethical standards of the International Coaching Federation — the same code of ethics that governs professional human coaches.
That means clear boundaries: coaching is not therapy, not counselling, and not medical advice. When a conversation moves beyond coaching, our coaches recognise it and respond appropriately rather than pressing on.
- Recognises when a topic is outside the scope of coaching
- Responds to signs of distress with a tested crisis protocol
- Signposts appropriate professional support
- Never diagnoses, prescribes, or advises on medical matters
- Holds ethical boundaries consistently — every session, every user
The standards we hold ourselves to
Safeguarding isn't only about how the product behaves in the moment. It's about the frameworks that shape how it's built.
UK and EU data protection law
We operate under UK and EU GDPR. Personal data is processed and stored within the European Union, and conversations are separated from identities. Our Data page covers this in full.
The EU AI Act
Our architecture is being shaped by the requirements of the EU AI Act as it phases in — designing for it now rather than retrofitting later, so safeguarding and compliance are built into the product from the ground up.
Our own quality framework
My Conversation Blueprint sets the safeguarding parameters every session is measured against — a standard we hold ourselves to on top of external codes, not instead of them.
When a conversation signals real risk
Alongside normal coaching, the session runs one immediate-risk safety override. It is triggered when the conversation signals a risk of serious harm — in three areas.
Self-harm
Language signalling that someone may be at risk of harming themselves.
Suicidal thoughts
Language signalling suicidal thoughts or intent.
Harm to another person
Language signalling a risk of harm to someone else.
What the override does
When it triggers, normal coaching stops. The coach responds directly and compassionately — it does not continue with goal-focused coaching — and encourages the person to reach their local emergency services, a local crisis resource, or a trusted person.
For users in the UK, that includes directing them to Samaritans on 116 123. Elsewhere, the coach uses location context to point to a relevant local crisis resource. This is an in-conversation response that signposts support — not a transfer, and not an automated alert to any third party.
Don't take our word for it
We've tested our safeguarding at length — but the fastest way to trust it is to try it. Open a guest session, say “I'm feeling really down…” and watch how the coach responds. Free, no account needed.
