Talkspace sees big opportunities to lead AI in mental health as chatbots draw scrutiny

Online therapy provider Talkspace reported another strong quarter with 25% revenue growth, driven by its expanding payer business, with net income of $3.3 million, up 73% from the same period in 2024.

The company brought in $59.4 million in revenue in the third quarter, driven by a 42% year-over-year increase in payer revenue, or insurance-covered sessions, to reach $45 million. Talkspace's direct-to-consumer business, however, continued to decline, with $4.6 million in revenue, down 23% year over year. The company's direct-to-enterprise revenue was $9.3 million, down 1% year on year.

The company reported adjusted EBITDA of $5 million, an improvement from $2.4 million adjusted EBITDA in the third quarter of 2024.

The company completed 432,000 insurance-covered mental health sessions in the third quarter, up 37% year over year, and active payer members increased 29% in the third quarter to 129,000, Ian Harris, Talkspace's chief financial officer, said during the company's third-quarter earnings call.

"This acceleration in the quarter is a reflection of the focused approach we initiated this year on all aspects of the patient journey," Talkspace CEO Jon Cohen told investors and analysts on the call.

The company's third-quarter results came in mixed relative to Wall Street expectations. Revenue for the quarter came in above the $57.4 million consensus target, while analysts expected adjusted EBITDA to come in at $5.4 million, according to William Blair analyst Ryan Daniels.

William Blair analysts "remain confident" that profit momentum will accelerate through the fourth quarter as the company's investments to convert covered lives into paying customers continue to bear fruit. 

"We view Talkspace as a compelling investment and anticipate significant expansion in the virtual behavioral health market in the coming years, as demand remains strong and patients of all ages (and payer types) continue to seek care," Daniels wrote in an analyst note. "We are also interested in hearing about investments in and commentary on the impact of AI on the behavioral industry, as well as updates on the TALK LLM market launch and monetization efforts. Lastly, we believe recent investor concerns about increased competitive entry into the space are overdone, as it will take time for direct-to-consumer operators to convert to more commercial sales."

Talkspace connects people via an app with therapists who provide counseling remotely, either over the phone, by video chat or by text. The company's suite of mental health services includes therapy for individuals, teens and couples as well as psychiatric treatment and medication management.

The company now covers nearly 200 million people in-network and through Medicare and Medicare Advantage plans as well as TRICARE, which serves U.S. military members and their families. Individuals also have access to Talkspace's services through employee assistance programs (EAPs), its partnerships with leading healthcare companies or as a free benefit through their employer, school or government agency. 

In the third quarter, Talkspace became in-network with several new Blues plans, including Illinois and Massachusetts, and the company won a competitive takeaway of one of the largest national EAPs, which we launched earlier this month. We have also made meaningful progress embedding Talkspace into our payer partner ecosystems by focusing on the areas that matter most to them," he said.

Payer revenue in the third quarter also benefited from strength in Talkspace's psychiatry business, Cohen said, which the company relaunched earlier this year to address the needs of high-acuity users and those who need medication. 

Talkspace narrowed its full-year guidance for 2025 and expects full-year revenue to be between $226 million and $230 million, representing year-over-year growth of 20% to 23%, Harris said. Talkspace forecasts adjusted EBITDA to be between $14 million and $16 million for 2025.

While Talkspace's growth is propelled by its expanding payer business, the company also is exploring opportunities in the pharma space to support patients. 

The company unveiled a partnership to support Novo Nordisk's WeGoTogether app, which is a tracking and support tool for patients taking weight-loss drug Wegovy. Talkspace's social health platform, Wisdo, which it acquired earlier in October, will provide users with group coaching and emotional support. Within the WeGoTogether app, Wisdo will power the group coaching experience that helps participants "build sustainable habits, share encouragement and stay emotionally supported as they work toward their health goals," Cohen said.

"This is a significant positive issue relative to the Wisdo go-to-market strategy. It was very clear that the people who are taking GLP-1s relative to weight management are really looking for a group coaching solution to help them get through it by seeing and interacting with other people who are going through the journey.

Talkspace is looking at "other similar opportunities in the pharma space," Cohen added in a Q&A with investors and analysts. "We need to get this one, honestly, right the first time around. So I would say more to come on that."

The company also is making significant investments in artificial intelligence, seeing opportunities to improve the experience for patients and cut down on paperwork for providers. 

"We continue to make significant progress integrating our work on AI into all aspects of our business with multiple initiatives to improve the customer member journey," Cohen noted, citing large language model (LLM) search engine optimization, AI assistance in improving eligibility determination for insurance and smart insights for providers in preparation of their sessions.

Other AI capabilities include comprehensive "smart evaluations" that give providers a HIPAA-compliant AI draft of a biopsychosocial evaluation after intake sessions, "smart notes" providing post-session summaries for the clinicians, the company's Talkcast individualized podcast and using AI to help review medical records to support compliance and clinical quality teams, he said.

The company is seeing value from those AI investments -- when providers use smart insights ahead of their second session with a member, those members are more than 30% more likely to book a third session within 30 days and 31% more likely to complete their third session within 30 days of registration, according to Cohen. The company's data also shows that 21% of people are more likely to book and complete a third session after listening to their personalized Talkcast podcast.

The company also launched three additional proprietary risk algorithms to add to its suicide risk algorithm. 

Talkspace also is building out a proprietary behavioral health AI model trained on its internal, de-identified clinical data sets, as the company claims to have the "largest behavioral health datasets in the industry," consisting of millions of therapeutic interactions on the Talkspace platform over the past 12 years. 

"Our proprietary LLM has been trained on hundreds of millions of anonymized therapy transcripts and rigorously tested for safety and therapeutic quality, and we envision it will serve as a therapy companion and clinical support tool," Cohen said. 

With the ongoing demand for mental health services, many individuals are turning to general-purpose LLMs and chatbots, such as ChatGPT, for emotional and mental health support, but many experts warn that they are not suitable for therapy. Media reports have revealed that youth have confided in AI chatbots and that has led them to isolation and even death. Congress is actively considering and introducing bipartisan legislation to regulate AI chatbots due to concerns over their significant potential harms, particularly to minors' mental health.

"[General-purpose chatbots] fail to challenge delusions or reinforce reality, lack real-time risk identification, lack clinical oversight and there is no HIPAA protection. We would note that even as companies work to address some of these issues, they do not have all of the necessary capabilities or experience to more fully protect users," Cohen said.

Talkspace wants to be a leader in offering safe, effective AI-supported mental health, given its 12 years of expertise in offering mental health services with strong clinical oversight and its AI model, backed by data and with a human in the loop, executives said.

The company can also refer an individual immediately to a live therapist, when needed, which is a critical feature most chatbots do not offer. 

"We recognize that these inadequacies of others is a unique opportunity for us. Our significant investment in an AI model, combined with our other capabilities, will offer individuals a significantly better and safer experience," Cohen told investors. "This new model that we have developed sets a new standard for both therapeutic efficacy and user safety, unlike general-purpose LLMs available today. Utilizing our database, the model was trained on hundreds of millions of tokens from anonymized and graded Talkspace-based therapy transcripts. In testing, the model consistently outperforms both open source baselines and state-of-the-art models in responding to high-risk mental health situations, including self-harm, hallucinations, OCD, mania and delusion."

In a recent test, Talkspace's LLM model, when compared to general-purpose models without specialized fine-tuning, delivered a 50% improvement in identifying and responding to high-risk behaviors, a 47% higher therapeutic quality score on the cognitive therapy rating scale and a 3x higher user satisfaction than the base model, he attested.

Talkspace's LLM model can also be used as an engagement tool for intake screening for multiple different types of patients and as an engagement tool in between sessions, he added.

The development of the Safe Talkspace AI agent's proprietary LLM is a large new opportunity with "significant potential upside for Talkspace's existing business and significant opportunity for new products," Cohen said.

The company expects to launch a full product offering in the first half of 2026. 

Cohen noted there are a number of "unique and significant commercial opportunities in the near term," as well.

"We expect to focus initially on several of these near-term opportunities that take advantage of our existing commercial infrastructure and brand presence, including as an affordable alternative for consumers and an attractive alternative for employers to provide a low-cost alternative to their employees," he said.

In response to a question from an analyst, Cohen said he sees opportunities to partner with AI and health tech companies to provide more rigorous clinician oversight for mental health support.

"I think we have a significant differential advantage relative to the other LLMs, specifically that we have the provider network. We actually know how to deal with mental health journeys. I think what's nuanced a little bit is if you're in another LLM and you need therapy and you need to see a therapist, we're a network, so what that means is if we have a continued seamless journey relative to the others, patients who need therapy essentially are going to be probably covered, so they don't have to pay more. If you're on one of the other LLMs and you need therapy or need to go outside the network, the question is who's going to pay for that," he said. "I do think there are licensing and other potential opportunities with other players out there."