United StatesHealthcare and life sciences2019 – ongoing

Talkspace

Proprietary AI for therapist matching and therapist insights

Talkspace said its AI matching raised third-session attendance by 50%

WorkingCompany's own claim

Problem

Talkspace wanted more new patients to stay in therapy past the first sessions, and wanted to catch risky behavior in patient messages.

Approach

Talkspace built its own AI on its therapy data, including a risk model, a therapist matching algorithm and smart insights shown to therapists. CEO Jon Cohen argued that general-purpose language models lack the capability to protect users in a mental health crisis.

Results

50%increase in patients attending a third session within 30 days after matching algorithm changes, per Talkspace's CEOCompany's own claim[1]

30%higher likelihood of booking a third session within 30 days of registration for members whose therapists use smart insights, per Talkspace's CEOCompany's own claim[1]

Takeaway

Both figures are the CEO's own and measure a third session within 30 days, not clinical outcomes.

Sources

Talkspace CEO Says General LLM Companies Lack Capacity to Address Mental Health Crises
Behavioral Health Business30 Oct 2025News article

Source summaries

[1]Behavioral Health Business30 Oct 2025

Behavioral Health Business reports Talkspace's CEO describing the company's AI built on millions of therapy messages and assessments. It repeats his claims of a 50% rise in third-session attendance from matching changes and a 30% higher chance of booking a third session with smart insights.

Full details are in the original sources.

Added 2 Nov 2026 · Updated 2 Nov 2026