Melbourne AI Scribe Startup Heidi Hits Near-Unicorn Status With Bold Pivot to Clinical Decision-Making Agents

Melbourne-based Heidi Health has vaulted into near-unicorn territory with a $480 million Series C funding round that values the AI medical scribe company at $1.27 billion. But the startup's ambitions now extend far beyond transcription: Heidi is pivoting toward "agentic" AI that can suggest clinical decisions and take supervised actions in patient care—a move that positions it at the forefront of healthcare AI's most promising and contentious frontier.

The funding round, led by Australian venture capital firm Blackbird with general partner Michael Tolo, reflects investor confidence in Heidi's rapid growth trajectory. The company has scaled revenue from $1.4 million to $71 million in just two years, with its AI transcription tool now used across 175 million patient encounters in 190 countries and supporting 110 languages.

From Scribe to Clinical Partner

Heidi's evolution represents a fundamental shift in how AI participates in healthcare delivery. Rather than simply documenting physician-patient conversations, the company is developing what co-founder and CEO Dr. Thomas Kelly describes as a "care partner" capable of recommending treatments and executing tasks under clinical supervision.

The company has built proprietary clinical AI models rather than relying on commercial frontier AI platforms—a strategic choice that may differentiate it from competitors using generic large language models. With the Royal Australian College of General Practitioners (RACGP) reporting that over 40% of GPs now use AI scribes, Heidi's market penetration demonstrates both the technology's appeal and the scale of its potential impact on clinical workflows.

Accuracy and Consent Challenges Mount

The company's ambitious expansion comes as research reveals troubling gaps in AI medical documentation quality and patient protections. A recent Australian study found variations in accuracy and thoroughness across AI scribes, including hallucinations—instances where AI generates plausible but false information. A separate evaluation of 7,545 AI-generated notes identified omissions, hallucinations, and accidental inclusions as common occurrences.

Ditto Transcripts CEO Ben Walker warned that AI-generated notes can be "cleanly written but incomplete," creating documentation that appears professional but may be "easy to overlook" despite containing clinically significant errors.

Privacy concerns add another layer of complexity. Some GPs have been recording consultations without explicit patient consent, a practice now banned under RACGP's updated general practice standards. Digital Rights Watch head of policy Tom Sulston has raised alarms about inadequate patient protections as AI adoption in clinical settings outpaces regulatory oversight.

Regulatory Crossroads for Clinical AI

Heidi's transition from passive documentation to active clinical decision support arrives at a critical moment for healthcare AI regulation. The company's near-unicorn valuation and rapid revenue growth validate the market opportunity, but the documented accuracy issues and consent gaps highlight the urgent need for oversight frameworks that can keep pace with technological advancement.

For healthcare providers, the promise of reduced administrative burden must be weighed against new liability concerns around AI-recommended treatments. For patients, the expansion of AI agents into clinical decision-making raises fundamental questions about consent, accuracy, and accountability—questions that current regulatory structures have yet to adequately address.

Sources

1. Australia's Heidi plans GP AI agents, nears unicorn status — ia.acs.org.au