HealthTechAIHealthcareSaaS

600 - Clinical Evidence at Your Fingertips: AI, Scribes, and the Future of Medical Documentation

Talking HealthTech
Talking HealthTech•Apr 22, 2026

Why It Matters

Heidi’s pivot demonstrates that practical AI‑driven documentation, not flashy triage tools, drives clinician adoption and reshapes revenue models in digital health.

Key Takeaways

  • •Early AI history‑taking tool struggled with practice workflow adoption.
  • •Multiple stakeholder resistance hindered Heidi’s original front‑door model.
  • •GPT‑4 outperformed in‑house AI, prompting product pivot to documentation.
  • •Template‑driven scribing achieved 90% weekly retention among clinicians.
  • •Free‑to‑pay scribe model leveraged cheaper models to drive adoption.

Summary

The Talking Health Tech episode chronicles Heidi Health’s journey from a 2020 AI‑driven history‑taking platform to today’s documentation‑focused scribe solution. Founder Thomas Kelly explains how the original "HX‑to‑DX" concept aimed to triage patients and generate differential diagnoses before the rise of large language models. Key insights include the product’s early promise hampered by complex stakeholder dynamics—practice managers, receptionists, and doctors all resisted a new asynchronous workflow. When GPT‑4 demonstrated superior unstructured‑data handling, Heidi pivoted, abandoning its custom AI engine in favor of off‑the‑shelf models that excelled at note generation and template customization. Kelly highlights concrete metrics: clinicians who adopted the templated scribe logged 90% weekly retention after just a few visits, and the team even released a free‑to‑pay version, betting that high‑frequency users would convert once model costs fell. He recalls the “history to diagnosis” branding and the painful realization that the AI itself was less valued than the streamlined documentation workflow. The shift underscores a broader industry trend: health‑tech firms must align AI capabilities with real‑world clinical processes, simplifying integration rather than over‑engineering novel workflows. Heidi’s experience suggests that pragmatic, low‑cost AI tools that enhance documentation can achieve rapid clinician adoption and sustainable revenue.

Original Description

In this episode of Talking HealthTech, Dr Max Mollenkopf is joined by Dr Tom Kelly to unpack the evolution of Heidi, an AI-powered clinical documentation tool, from its early experimentation in general practice to its growing presence across health systems globally.
The conversation traces the realities of building in digital health, including early missteps, product pivots, and the challenge of finding meaningful product-market fit in a complex and highly regulated environment.
The discussion goes beyond surface-level AI hype to examine how tools like Heidi are being shaped by real clinical workflows.
It explores the practical challenges of integrating with existing practice management systems, navigating regulatory frameworks across different regions, and competing in an increasingly crowded international market.
There is also a focus on the rise of AI-powered clinical scribing, the role of large language models, and how these technologies are changing the day-to-day experience of clinicians.
Alongside the technical and commercial considerations, the episode reflects on broader questions facing the sector, including safety, interoperability, and the cost of scaling AI in healthcare.
It also looks ahead to emerging use cases such as real-time evidence retrieval and patient-facing applications, offering a grounded perspective on where AI is delivering value today and where it may have the greatest impact in the future.
Key Takeaways
✨ Product evolution and lessons learned Heidi started as a triage and workflow automation tool and shifted focus to clinical AI scribing in response to market needs and advancements in AI models.
💬 Integration and interoperability challenges Integrating with practice management systems remains complex and highly dependent on regional vendors and technical standards, highlighting the ongoing struggle for efficient interoperability in healthcare.
🌍 Global competition and expansion Heidi's growth strategy and competition vary by country, with different primary competitors in Australia, the UK, and North America, and a keen focus on clinician-centric product development.
🛠️ Regulation and medical device territory AI-based clinical tools like Heidi must constantly reassess their compliance with evolving software as a medical device rules, which differ across international markets.
🤝 Feedback-driven design Heidi’s freemium model enabled rapid feedback loops from clinicians, shaping product features and spurring adoption in both individual practices and enterprise hospital deals.
Timestamps
00:00 Heidi’s origins and early pivots
03:37 Traction and barriers to product adoption
09:01 Pivot to AI-powered scribing
14:11 Drivers of Heidi’s growth and market position
18:56 Competitors in each market
22:45 Evidence product launch and vision
29:17 Software as a medical device, regulatory lines
37:37 Relationships with PMS providers
44:16 Interoperability barriers and national health records
52:13 The cost of AI in clinical tools
54:40 Expanding to communications and voice products
57:32 Consumer-facing applications and future directions
59:59 Will AI replace clinicians?
--------
Want to keep the conversation going?
In the THT+ Community, our members keep the discussion going, share perspectives, and unpack what this episode actually means in practice with other people working in healthtech.
If you’re enjoying the show and want access to full articles, episode discussions, meetups, and member-only content, you can learn more about becoming a THT+ Solo Member here: https://talkinghealth.tech/solo_shownotes.
And if this episode was useful, leaving a review or sharing it with someone in the industry always helps.

Comments

Want to join the conversation?

Loading comments...