Mental health and wellbeing
Serena
A WhatsApp companion for emotional listening and support, for women with few places to turn.
- Sector
- Mental health and wellbeing
- Stack
- WhatsApp Business APILanguage modelsNode.jsConversation analytics
Active listening, validation and containment over WhatsApp. The most constrained system we have built, because here the failure modes are harm.
The people this serves often have no counsellor and no privacy to speak to one. What they have is WhatsApp, on a phone, at two in the morning.
The service listens. It does not advise, diagnose or tell anyone how to feel. When a situation exceeds what listening should handle, it says so and routes to professionals.
Risk detection is the feature everything else serves
An active crisis triggers an immediate handoff to a national helpline. Moderate risk is held in conversation with the helpline offered alongside, not instead.
Validation without normalisation
Saying it is normal to feel this way quietly minimises the experience. The system is steered away from that whole class of reply.
Sounding like a person, not a product
Lists and polished paragraphs read as a machine processing a case. Responses are short and plain, like someone typing on a phone.
Meeting people inside their real constraints
Gender roles, economic dependence and absent support networks shape what is possible. The service explores what is actually available rather than reciting a script.
Measuring a service you cannot read
Conversations are confidential, so transcripts are off limits. Quality comes from aggregate signals: timing, duration, endings, handoffs.
Tell us where you are starting from. We take it from there.
Start a project