Nelson Advisors: Heidi Health, Best Practice and the challenge facing clinical AI is not accuracy of ML models, but access to data infrastructure and integration permissions within established systems
The Genesis of the Turf War: From Ambient Scribes to Autonomous Agents
The primary care health technology landscape is undergoing a structural transition from assistive ambient transcription utilities to autonomous, agentic artificial intelligence systems. For several years, ambient voice technology (AVT) operated as a non-invasive layer at the clinical periphery. These early tools ingested consultation dialogue, generated structured consultation summaries, and relied on general practitioners to manually copy and paste text into their local Electronic Medical Record (EMR) software. However, the economic and operational value within digital health has aggressively migrated from basic documentation to workflow automation, intelligent decision orchestration, and the autonomous execution of administrative tasks.
This technological migration reached an inflection point with the release of Heidi II by Melbourne-based digital health company Heidi Health. Valued at approximately $1.26 billion following venture capital funding rounds led by Blackbird Ventures, Heidi positioned its next-generation platform not merely as a documentation aide, but as an agentic partner capable of executing clinical administration without direct minute-by-minute manual guidance. Operating across Australia, New Zealand, North America, and the United Kingdom, Heidi designed its updated platform to navigate clinical software interfaces, extract longitudinal patient context, and manage pre- and post-consultation correspondence.
The commercial deployment of these autonomous capabilities provoked an immediate confrontation with Best Practice Software, the entrenched provider of Australia's most widely utilized general practice software suite, Bp Premier. Following the launch of Heidi II, Best Practice issued a critical communication to its client base, warning medical clinics that running Heidi II in conjunction with Bp Premier represents a direct violation of its End User Licence Agreement (EULA). Best Practice's intervention highlights an intensifying dispute regarding system boundaries, data sovereignty, and control of the primary care clinical interface.
The Gatekeeper’s Posture: Contractual Breaches, Data Integrity, and Safety Allegations
Best Practice Chief Executive Craig Hodges formalised the vendor’s resistance by issuing an urgent compliance directive directly to general practitioners and practice managers. The company grounded its prohibition on three central arguments: clinical governance and data integrity, statutory privacy compliance, and medical device regulatory exposure.
Hodges informed users that Heidi II was apparently designed to operate within Bp Premier through automated interface navigation and process automation. Best Practice asserted that this unauthorized operational profile introduced grave risks to the clinical record, claiming the software could alter or delete patient medical histories, clinical observations, and diagnostic test results without explicit, contemporaneous clinician oversight.
From a contractual standpoint, Best Practice emphasized that its EULA explicitly prohibits licensees from allowing third-party applications to access, interact with, or modify its database or core software without prior written consent. To reinforce its legal barrier, Best Practice announced an immediate revision of its licensing terms in the Oxford Service Pack 2 (SP2) update for Bp Premier, formally barring the use of robotic process automation (RPA), scraping tools, scripts, and autonomous bots within its environment.
Best Practice compounded these contractual warnings by raising compliance concerns under the Privacy Act, arguing that automated tools interrogating databases without documented patient consent expose practice principals to legal liability. Hodges further invoked emerging regulatory guidance, warning that tools performing autonomous clinical actions risk classification as regulated medical devices under Australian therapeutic goods legislation.
Analytical Vector | Best Practice Software Legal & Governance Posture | Heidi Health Commercial & Technical Defense |
Licence Compliance | Strict enforcement of EULA; bans all unapproved bots, screen scrapers, and automated third-party tools. | Characterizes terms as anti-competitive gatekeeping designed to restrict clinical choice and market access. |
System Interaction | Highlights risks of unmonitored software altering, corrupting, or deleting clinical records. | Operates via supervised user interface automation without writing directly to the underlying database. |
Clinical Governance | Asserts that autonomous background actions degrade patient safety by bypassing contemporaneous GP sign-off. | Uses a default "Manual Mode" where the clinician inspects and approves all workflows and summaries. |
Privacy Safeguards | Claims data extraction without bespoke, explicit consent creates severe exposure under the Privacy Act. | Maintains ISO 27001, SOC 2 Type II, and Australian Privacy Principles compliance with audited data sovereignty. |
Regulatory Exposure | Warns that agentic systems crossing into clinical analysis operate as uncertified medical devices. | Maintains ongoing dialogue with the TGA and asserts compliance with international digital health standards. |
Defensive Moats and Ecosystem Keiretsu: Sonic Healthcare and the Lyrebird Alliance
Understanding Best Practice’s aggressive perimeter defense requires examining the concentrated ownership structure of the Australian primary care software market. Best Practice holds an estimated 80% market share across Australian general practice clinics, giving Bp Premier a de facto monopoly over primary care IT infrastructure. With legacy rivals such as Telstra Health’s MedicalDirector steadily ceding territory on an aging architecture, Bp Premier serves as the essential distribution channel and operational choke point for digital primary care delivery.
This market position is reinforced by the strategic ownership web of Sonic Healthcare, the ASX-listed pathology and radiology conglomerate that holds a 49% stake in Best Practice. Sonic’s primary care footprint operates as a healthcare keiretsu—an integrated network of aligned commercial interests that connects pathology laboratories, diagnostic imaging centers, corporate medical clinics through the Independent Practitioner Network (IPN), and practice analytics via a 51% controlling interest in Cubiko. Within this structure, desktop software serves as a protective distribution channel that secures and routes high-margin pathology and diagnostic referrals back to Sonic's laboratory network.
The emergence of independent artificial intelligence platforms threatened to disrupt this ecosystem by establishing an external control layer over clinical workflows. Best Practice responded by adopting a twin strategy of platform containment and internal alignment. Rather than maintaining vendor neutrality, Best Practice made a direct equity investment of $3.378 million into Lyrebird Health, a domestic AI scribe competitor. Alongside this investment, Best Practice engineered an exclusive, native product integration that embeds Lyrebird directly into the Bp Premier desktop toolbar.
To curb the rapid adoption of Heidi Health, Best Practice introduced an exclusive zero-cost tier of Lyrebird Scribe embedded natively within the core Bp Premier application. Unlike external competitors, Lyrebird was granted deep, bidirectional application programming interface (API) access to the Bp Premier database. This privileged integration enables Lyrebird to read patient records, active medication profiles, and historical notes directly, while autonomously writing finished progress notes, clinical measurements, and complex Chronic Disease Management Care Plans (MBS Items 721 and 732) back into the system.
By granting native write-back privileges exclusively to an entity in which it holds an equity stake while legally barring third-party competitors under the banner of data safety, Best Practice effectively leveraged its gatekeeping power to insulate its desktop environment from venture-backed market entrants.
The Technical Workaround, Interoperability, and the Agentic Paradox
Heidi Health Chief Executive Dr. Thomas Kelly directly refuted allegations that Heidi II was built to compromise clinical integrity. Kelly stated that Heidi had spent three years attempting to establish formal, structured integration channels with Best Practice, but was repeatedly met with commercial resistance and an opaque partner program that withheld production API credentials. Confronted with an uncooperative incumbent, Heidi chose to bypass the API gatekeeper altogether by engineering desktop-level automation.
Rather than injecting data directly into the back-end SQL database, Heidi II functions by emulating human clinician interaction. The agent navigates the local computer interface, simulating mouse and keyboard inputs to read chart histories, review specialist letters, and extract context directly from the displayed screen. Kelly explained that the platform was designed with strict operational guardrails that explicitly prevent the system from committing write actions, generating diagnoses, or ordering investigative tests without direct user interaction. The system defaults to a manual mode where every extracted data point and drafted document requires affirmative clinician sign-off, reflecting the supervision model used with junior medical registrars.
Heidi framed the dispute around data ownership and clinician autonomy. From Heidi’s perspective, the electronic health record belongs ethically and legally to the medical practice, the consulting physician, and the patient—not to the software vendor that licenses the database architecture. Kelly argued that using end-user licensing agreements to dictate which cognitive tools a medical practitioner can run on their personal workstation represents an anti-competitive restraint of practice.
However, Heidi’s push for open interoperability faced scrutiny when clinical users examined the startup’s own legal documentation. Newcastle general practitioner Dr. Max Mollenkopf publicly highlighted Clause 7.2.9 in Heidi Health’s standard Master Services Agreement. The clause explicitly prohibited Heidi's own customers from utilising automated tools, scrapers, scripts, or bots to access or interact with the Heidi platform without express API authorisation, the very technical restriction Best Practice was enforcing against Heidi.
Heidi’s executive team acknowledged the contradiction. Dr. Simon Kos, Heidi’s Global Chief Medical Officer and former Microsoft digital health executive, alongside Dr. Kelly, conceded that the clause was a standard cloud rate-limiting provision that failed to reflect the company's pro-interoperability principles. Heidi committed to redrafting the clause, but the episode highlighted the tensions software vendors face when balancing platform security against open integration demands.
Regulatory Realities: The TGA Digital Scribe Review as Strategic Leverage
The clash between Heidi and Best Practice unfolded alongside increased regulatory scrutiny from the Therapeutic Goods Administration (TGA). On 16 October 2026, the TGA released the outcomes of its comprehensive market review into digital scribing software supplied across Australia, reshaping the compliance obligations governing AI developers and healthcare providers.
The regulator investigated twelve representative digital scribing applications operating in the domestic market. While none of the twelve software suppliers had self-identified their platforms as medical devices, the TGA determined that eight of the twelve products contained integrated features that likely satisfied the statutory definition of a medical device under Section 41BD of the Therapeutic Goods Act 1989.
The TGA established a firm regulatory division based on software functionality rather than vendor disclaimers. Software that strictly limits its operations to recording, transcribing, and formatting clinical conversations into notes or letters for human review remains classified as an administrative tool exempt from device registration. Conversely, systems that analyze clinical conversations to offer unprompted diagnostic suggestions, formulate clinical risk scores, highlight red flags, or automate clinical investigations are legally categorised as Software as a Medical Device (SaMD) requiring formal inclusion on the Australian Register of Therapeutic Goods (ARTG).
Clinical Software Functionality | Statutory Classification | Regulatory Burden & Mandatory Obligations |
Ambient Audio Transcription | Administrative Software (Non-Device) | Compliance with Australian Privacy Principles; no ARTG registration required. |
Clinical Progress Note Formatting | Administrative Software (Non-Device) | Requires contemporaneous human review prior to clinical finalisation. |
Diagnostic & Differential Generation | Regulated Medical Device (Class I/II SaMD) | Mandatory inclusion on the ARTG; pre-market conformity assessment. |
Clinical Decision Support (CDSS) | Conditionally Exempt Medical Device | Must explicitly display underlying clinical logic, citations, and guidelines to the user. |
Autonomous Chart Execution & Order Routing | High-Risk Medical Device (SaMD) | Excluded from CDSS exemptions; subject to clinical safety audits and adverse incident reporting. |
The TGA also tightened the Clinical Decision Support System (CDSS) exemption framework, requiring that any exempt software explicitly disclose its clinical logic, medical citations, and underlying guidelines during the consult to allow practitioners to verify recommendations.
Best Practice leveraged the TGA’s findings to support its stance against Heidi II. By highlighting the regulator's position that advanced generative AI tools routinely drift into regulated medical device classifications, Hodges warned that deploying uncertified autonomous workflow agents created direct regulatory exposure for clinics.
Heidi countered that its core documentation engine was already registered and validated under medical device guidelines in the United Kingdom, asserting that Heidi II operates strictly as an administrative efficiency tool rather than a diagnostic decision engine. Nevertheless, the TGA’s clarified framework increased the regulatory stakes for practice principals, who remain personally and legally responsible for ensuring that all software deployed within their clinics complies with the Therapeutic Goods Act and the Privacy Act.
Macro Implications: The AVT Consolidation and the Battle for the Primary Care Interface
The confrontation between Heidi Health and Best Practice reflects a major structural transition occurring across primary care infrastructure worldwide. Ambient voice and artificial intelligence technologies are driving an architectural shift that decouples the clinical interface from the transactional database.
Throughout the past three decades, the practice management system served as both the primary system of record and the exclusive user interface for general practitioners. Clinicians spent their working days trapped within complex, click-heavy interfaces designed primarily for billing, coding, and record storage. The rise of agentic AI platforms fundamentally alters this paradigm. As ambient systems demonstrate the capacity to listen, extract context, generate documentation, and automate administrative tasks, the AI layer becomes the primary interface where clinical reasoning and administrative coordination occur.
Under this model, the underlying EMR risks being relegated to a commoditized, passive data repository. Incumbent vendors whose platforms rely on aging client-server architectures face a clear commercial threat: if an intelligent agentic overlay captures the clinician’s attention and automates the workflow, the legacy EMR loses its ability to monetise third-party integrations, direct diagnostic referral pathways, and command software licensing premiums.
Best Practice’s aggressive EULA enforcement, strategic equity investments, and distribution subsidies through Lyrebird illustrate an incumbent playbook designed to prevent platform disintermediation. Unable to match the development velocity of venture-backed AI startups, legacy vendors utilise contractual control and vertical diagnostic relationships to defend their installed base.
Strategic Vector | Proprietary Closed EMR Monolith (Best Practice / Sonic) | Autonomous Intelligent Overlay (Heidi Health) |
Platform Architecture | On-premise, client-server legacy desktop deployment with centralized relational databases. | Cloud-native, micro service architecture operating across local desktop automation clients. |
Interface Control | Proprietary desktop interface; workflows tied to internal menus and closed toolbars. | Dynamic ambient overlay; clinical interaction occurs through natural language and automated UI navigation. |
Commercial Strategy | Defensive ecosystem; leverages software dominance to channel diagnostic and billing revenues. | Rapid feature iteration, platform disruption, and aggressive enterprise user acquisition. |
Integration Method | Walled garden; selective access granted only to equity partners and curated vendors. | Open API advocacy, multi-platform integrations, and UI-level robotic process automation. |
Vulnerability | Significant technical debt, platform inertia, and vulnerability to antitrust or interoperability mandates. | Fragility of UI automation, ongoing EULA litigation risks, and reliance on uncooperative host software. |
This conflict exposes three major operational and systemic risks across the digital health ecosystem:
The technical fragility of robotic process automation represents an ongoing vulnerability for modern AI startups attempting to bypass legacy gatekeepers. Relying on simulated keyboard strokes, visual coordinate mapping, and screen scraping allows vendors to bypass closed APIs, but leaves tools susceptible to unexpected failures. Standard software updates, slight interface revisions, or operating system patches released by the host EMR can alter interface hooks, break automation routines, and disrupt clinical practice administration. In a high-stakes clinical setting, such technical failures can lead to documentation errors and administrative bottlenecks.
The legal and clinical liability burden has been shifted directly onto primary care practices. By issuing a formal alert declaring Heidi II a licence violation and warning of potential privacy breaches, Best Practice placed medical practice owners in legal jeopardy. If a clinic runs unauthorized automation software and subsequently suffers an administrative error, privacy breach, or adverse clinical event, the practice risks voiding its vendor support contracts and complicating its professional indemnity insurance protections. As a result, clinic principals are forced to choose between the operational efficiency of agentic AI and contractual safety with their primary EMR provider.
The escalation of these disputes increases the likelihood of regulatory and antitrust intervention regarding healthcare interoperability. The Australian general practice software market—where a single incumbent holds 80% market share and utilizes contractual restrictions to insulate its own venture investments against outside competition—highlights the limits of voluntary data sharing. When incumbent vendors block data access to defend their commercial models, patient care continuity and technological innovation suffer.
Without government-mandated open API standards, secure FHIR specifications, and explicit statutory data-blocking prohibitions similar to those enacted in other advanced jurisdictions, entrenched vendors will continue to serve as commercial gatekeepers.
Strategic Outlook and Conclusions
The dispute between Heidi Health and Best Practice Software is an early battle in a broader market restructuring. As artificial intelligence transitions from passive documentation transcription into active workflow automation, the competitive center of digital healthcare is shifting. The dispute over Heidi II demonstrates that the primary challenge facing clinical AI is no longer the accuracy of machine learning models, but access to data infrastructure and integration permissions within established systems.
Incumbent software providers will likely continue defending their desktop market share using licensing restrictions, closed partner integrations, and vertical alliances with diagnostic and clinical service conglomerates. In response, venture-backed digital health startups will continue to rely on desktop automation, rapid feature development, and clinician-led adoption to apply pressure on closed ecosystems.
However, robotic process automation remains an unstable foundation for critical national health infrastructure. Bypassing API barriers through user-interface emulation introduces operational fragility and governance challenges in high-stakes clinical settings.
Ultimately, this conflict illustrates that genuine primary care transformation cannot occur within walled gardens. Until national health policy establishes enforceable data interoperability standards that grant clinics true ownership of their health data, general practitioners will remain caught between legacy software gatekeepers defending their territory and venture-backed AI platforms attempting to reshape clinical administration from the outside.
Nelson Advisors > European Healthcare Technology Investment Banking
Nelson Advisors specialise in Mergers and Acquisitions for European HealthTech, MedTech, Digital Health, Healthcare IT, Healthcare AI companies in the Lower to Mid Market ranging from $25M to $250M EV. www.nelsonadvisors.co.uk
Healthcare.Digital is the Google News approved HealthTech and MedTech Thought Leadership platform for Nelson Advisors, positioning them as a specialised authority on European Healthcare Technology M&A and strategic corporate development. https://www.healthcare.digital
Nelson Advisors publish Europe's Leading Healthcare Technology Investment Banking Newsletter every week, join 5000+ HealthTech and MedTech subscribers today! https://lnkd.in/e5hTp_xb
Nelson Advisors LLP
Hale House, 76-78 Portland Place, Marylebone, London, W1B 1NT
Meet Nelson Advisors @ Events in 2026
Digital Health Rewired > March 2026 > Birmingham, UK
NHS ConfedExpo > June 2026 > Manchester, UK
HLTH Europe > June 2026, Amsterdam, Netherlands
HIMSS AI in Healthcare > July 2026, New York, USA
Bits & Pretzels > September 2026, Munich, Germany
HealthInvestor Healthcare Summit > September 2026, London, UK
World Health Summit 2026 > October 2026, Berlin, Germany
HLTH USA 2026 > October 2026, USA
Global Health Exhibition 2026 > October 2026, Riyadh, Saudi Arabia
Web Summit 2026 > November 2026, Lisbon, Portugal
MEDICA 2026 > November 2026, Düsseldorf, Germany
Leaders in Health Summit 2026 > November 2026, London, UK
Venture Capital World Summit > December 2026, Toronto, Canada
Meet Nelson Advisors @ Events in 2027
Digital Health Rewired > March 2027 > Birmingham, UK
Barclays Health Elevate > March 2027, London, UK
NHS ConfedExpo > June 2027 > Manchester, UK
HLTH Europe > June 2027, Amsterdam, Netherlands
Comments