Interview

Fortuna Health raises Series A to build TurboTax for Medicaid enrollment

Jul 21, 2025 with Nikita Singareddy

Key Points

  • Fortuna Health closes Series A led by investor Jason to scale a B2B2C platform that simplifies Medicaid enrollment across 56 fragmented state jurisdictions, each with different rules and submission formats.
  • The company distributes through health plans and hospitals rather than direct-to-consumer, embedding enrollment tools into member communications and hospital intake workflows to absorb customer acquisition costs.
  • General AI agents solve only 30% of the problem; Fortuna's defensibility comes from integrations with state systems and mapped workflows for edge cases like fax-based submissions and account splits that remain widespread across Medicaid programs.
Fortuna Health raises Series A to build TurboTax for Medicaid enrollment

Summary

Fortuna Health has closed a Series A round led by Jason (firm name not disclosed in segment) to scale its Medicaid enrollment and renewal platform. Nikita Single, co-founder and CEO, describes the product as "TurboTax for Medicaid" — software that guides users step by step through a fragmented, state-by-state enrollment process that has existed largely unsupported by consumer technology despite covering one of the largest public health programs in the country.

The company launched roughly two and a half years ago. The core observation driving the business is structural: every U.S. state runs its own Medicaid program under a different name, with different rules, different submission formats, and in many cases different managed-care insurers (United, Centene, Elevance/Anthem) contracted to administer benefits. That complexity creates a meaningful coordination and navigation problem that no scaled consumer product had addressed.

Business model and go-to-market

Fortuna operates B2B-to-consumer, not direct-to-consumer. Its paying customers are the managed Medicaid health plans and hospital systems — not individual enrollees. Health plans distribute Fortuna via text or email to members who need to enroll or renew, embedding the product directly into existing member communications. Hospitals use it as an enrollment assist for uninsured patients presenting for care. Customer acquisition cost is effectively absorbed by the institutional partner.

The company is rolling out state by state, building density in markets like New York before expanding. The full addressable footprint is 56 jurisdictions — 50 states plus territories including Guam.

Why the problem is harder than it looks

Single pushes back on the assumption that a general-purpose AI agent could commoditize Fortuna's core function. Her estimate is that a tool like ChatGPT solves roughly 30% of the problem — the portion involving basic rules and website navigation. The remaining complexity is operational and real-world: forgotten passwords from accounts created years earlier, account splits when a dependent ages off a parent's coverage, job changes that must be reported to state systems via fax lines not publicly listed anywhere. Fortuna's defensibility lies in its integrations with state systems and its accumulated mapping of those edge-case workflows, which compound quickly into the majority of actual user scenarios.

Fax machines remain a live submission channel across multiple state Medicaid programs, in some cases because government procurement rules prohibit contracting with vendors outside long-standing Xerox-era agreements. Fortuna automates across fax, snail mail, digital portals, and email depending on what each state program accepts.

Broader healthcare AI view

Single flags EHR-to-insurance data conversion — the translation layer between how physicians write clinical notes and how insurers process them — as a sector with substantial unsolved opportunity. She also identifies voice AI for clinical settings as early-stage and underdeveloped, and expresses strong conviction in AI nurses for low-acuity care as a structural response to ongoing physician and nursing shortages, with or without a human review layer.