Maven Clinic has introduced Maven Intelligence, an artificial intelligence-powered infrastructure embedded across its virtual clinic, care programs and benefits platform.
The company refers to Maven Intelligence as an orchestration layer that integrates agentic AI with longitudinal data and clinical expertise to personalize care. The layer is built on over 1 billion structured data points from Maven care journeys and will begin rolling out to members in March.
“Women’s health has [long been] a major gap in healthcare,” Jaya Savkar, senior vice president of product at Maven, told Fierce Healthcare. “We really believe that we have an opportunity to help close the gap … We’re the largest virtual clinic for women and families, and by coupling that with women’s health-focused AI, we believe that humans and AI can do more together.”
Maven’s conversational AI is trained to take women’s health symptoms seriously and support the clinical and emotional nuance of care journeys. Maven Intelligence is based on OpenAI and Google's large language models (LLMs) via business associate agreements. As a result, the tool is HIPAA-compliant and does not pull insights from the entire internet. Instead, the closed model draws on a Maven member’s history, goals, benefits coverage and integrated data. This includes medical records, lab results and wearables.
The tool can offer personalized, evidence-based guidance, connect members to the right providers, navigate benefits and prompt follow-ups or adherence support. Every interaction feeds back into the AI’s guidance based on what demonstrably improved aggregated, program-level clinical outcomes, such as natural conception rates, NICU admissions and C-section rates.
“Ultimately, we want, for example, members to never have to repeat themselves,” Savkar said. “We want providers to have all the latest information about a member, so that when a member interacts with them, the provider already knows their history and they don’t explain that again.”
As part of its safety architecture, Maven is leveraging Nvidia’s NeMo Guardrails, an open-source toolkit for easily adding programmable guardrails to LLM-based conversational systems.
“That helps ensure that conversations stay on appropriate topics, [and] they follow safe and clinically-informed guidelines,” Savkar said.
Maven Intelligence underwent multidisciplinary review and performance evaluation alongside clinicians. Maven runs automated evaluations to ensure the system is consistently addressing member questions. A human team also manually reviews a sample of conversations on an ongoing basis to check for bias or gaps. That feedback is used to improve the system. Conversations are not sold or used to train external models.
AI is already embedded across the Maven platform, supporting care navigation, clinician workflows and member engagement. So far, these capabilities have helped accelerate access to care, surface earlier risk signals and reduce administrative burden, Maven said. Specifically, AI has helped cut post-appointment documentation time by 30%, per the company.
“What we see is that AI is not replacing humans, but we really believe it is complementing them so that by the time a member is talking to a provider, they can already have good information,” Savkar said.