Talkspace, Inc.TALK

Talkspace has a proprietary AI agent in active beta set to launch later this year.

“Dr. Jon Cohen, CEO of Talkspace, said, “Talkspace concluded 2025 with strong momentum, driven by a record fourth quarter where we successfully prioritized network curation, product innovation, and deeper payor integrations. These strategic initiatives delivered 29% year-over-year growth with $63.0 million quarterly revenue and a 22% year-over-year increase in total annual revenue to $228.9 million. With our proprietary AI agent now in active beta and set to launch later this year, we are well-positioned to deliver sustainable growth and long-term value in 2026.””

Talkspace has developed a specialized safe AI agent for mental health support.

“The Talkspace AI Agent 12 2026 JP MORGAN HEALTH CONFERENCE General-purpose LLMs now utilized by 10% of the world’s population (800M+ people) due to their always -available, low-cost, anonymous, and easy-to-access nature – making them a default option for mental health conversations While these bots have “democratized” access, they were never built to support mental health, leading to a rash of reported harmful outcomes Mental health support must be far more specialized and nuanced including: – Challenge distorted thinking – Recognize delusions – Identify risk in real time Designed to be the first “safe” AI agent specifically designed for mental health support and developed utilizing clinically recognized standards of care with continuous human oversight, and privacy (HIPAA) protection Trained and fine-tuned on Talkspace’s massive mental health dataset to improve risk recognition, support appropriate decision-making, and avoid the pitfalls already seen in general-purpose AI systems Keeps clinicians constantly in the loop with clear escalation pathways to connect users at risk to a licensed human clinician if needed Does not replace clinicians but extends their reach, adhering to strict clinical standards while identifying millions of new users who require human intervention”

Talkspace promotes its purpose-built mental health AI agent for consumer adoption.

“13 2026 JP MORGAN HEALTH CONFERENCE What it is How to use it What it is for A private guide that leads you toward answers (and never follows you down the wrong path) Share fears or insights anytime Your space to work things out and get clarity Confide without judgment or time limits Fast support and lasting progress Start changing your life today It keeps your thoughts completely private, unlike any AI Choose the AI guide that’s purpose-built for mental health and backed by science Turn to it for trusted, science-backed guidance Talk to it to check reality, stop spirals, or avoid burnout Vent to it instead of blowing up the group chat Work on yourself, all by yourself Oversharing with other AIs? Switch to this for guidance you can trust For anyone who wants to make progress on their mental health starting this second For those who want clarity without judgment For those searching for answers in the wrong places For those whose friends have stopped listening Positioning: Driving Consumer AI Agent Adoption”

Talkspace describes its first purpose-built mental health LLM and AI agent.

“The First Purpose-Built Mental Health LLM 16 2026 JPMORGAN HEALTHCARE CONFERENCE A next-generation mental health solution powered by a LLM trained on real clinical data, evidence-based care, and aligned safety standards Our Breakthrough AI Agent Trained on one of the world’s largest MH databases (10B words, millions of transcripts) Patient intake and consent HIPAA-protected, patient-consented environment Proprietary risk algorithms embedded Continuous monitoring by professional therapists Escalation process for higher levels of care Talkspace AI Agent: Personalized, Clinically Grounded”

Talkspace details its clinical databank used to build its AI agent.

“Built on a Clinical Databank Powered by Millions of Patient Journeys 17 2026 JP MORGAN HEALTH CONFERENCE 8 Billion words exchanged 140M messages 300M tokens from real session 1.2M clinical diagnoses 4.3M psych notes 3M therapist ratings 450K treatment plans 1.4M treatment goals 3.8M clinical interventions 12+ years longitudinal, de-identified under HIPAA Safe Harbor / Expert Determination Approach to Building our AI Agent Built on Talkspace expertise, proprietary datasets, and evidence-based guardrails Evaluated with clear benchmarks, rubrics, and validated clinical scales Developed transparently with a safety-first approach”

Talkspace outlines safety, clinical oversight, and escalation for its AI agent.

“Built Prioritizing Safety and Clinical Oversight 18 2026 JP MORGAN HEALTH CONFERENCE Operates fully within Talkspace’s QMS – every AI output is validated by licensed clinicians, documented under QMS control, and transitioned to human care when needed to ensure clinical safety and trust AI Agent Intake Risk assessment Live therapist AI Agent Safe Escalation Workflow: HIPAA Compliant Safety & quality model Risk algorithm Clinical oversight Consent Escalation”