Detect mental health risk in AI Chatbots before it becomes a crisis.

Book a Demo

Wondi is a clinician validated API, that provides a trajectory based mental health risk scoring for AI companion platforms, built for millions of conversations.

Clinician Validated

Built on real clinical science. By real clinicians.

5 years of MIT longitudinal research. Peer reviewed methodology. A psychiatrist on the founding team. Wondi is the only mental health risk tool for AI conversations built from clinical ground up.

99%

Clinical Classification Accuracy

5 Years

MIT Longitudinal Mental Health Research

Psychiatrist Led

Clinical Methodology Built By Mental Health Experts

Book a Demo

Early Detection

Risk scores that read where the conversation is going.

Wondi tracks conversation trajectories in real time. Not single messages. Not keywords. The full arc. So you intervene before someone reaches the danger zone.

Trajectory Scoring

Risk updates continuously as the conversation evolves.

Early Intervention

Redirect conversations before they reach the danger zone.

Full Audit Log

Every signal scored, timestamped and documented.

Compliance Ready

The law is already here. Wondi keeps you ahead of it.

California SB 243 is live. 11 more states are following. Wondi gives you the evidence based protocols regulators require and the audit trail to prove it.

California SB 243

AAIDEN Act | Live January 1, 2026

Requires evidence based mental health crisis protocols for AI companion platforms.

New York Article 47

S.3008 | Live November 5, 2025

Mandates crisis identification protocols and disclosures every 3 hours for AI companions.

Illinois WOPR Act

HB 1806 | Signed August 2025

Unanimously passed. AI cannot independently deliver mental health care without licensed professional oversight.

How it works

We read linguistic patterns. Not personal data.

One API call per conversation. Nothing stored. Nothing retained.

Privacy by design

No user IDs, no personal information, no sensitive data sent anywhere. Wondi analyzes linguistic patterns in the conversation text only. Data processing agreement included as standard.

Minimal lift to get live.

A single API call per conversation. A risk score comes back in milliseconds. No changes to your existing infrastructure, no new systems to maintain.

You stay in control.

Set your own thresholds. Decide what happens with the score. Wondi gives you the signal. What your platform does with it is entirely up to you.

Get in touch

Your platform is talking to people in distress right now.

Book a Demo

Find out what Wondi sees in your conversations.