Healthcare has always advanced on the back of new tools, from the stethoscope in 1816 to the MRI in 1977, but each wave also added new strain on the workforce through more clerical, cognitive and coordination burdens.
Today’s labour challenges in healthcare may be urgent, but they are not new. For decades, labour demand has run ahead of supply. The WHO projects a global shortfall of 11 million health workers by 2030. Almost half of US physicians report burnout, much of it linked to clerical burden and after-hours work catching up on electronic health records, typing notes, entering orders and dealing with the administrative side of digital systems rather than direct patient care.
These are all symptoms of a system that cannot scale with people alone.
AI is the first technology in decades with the potential to take work away.
The first success stories are already emerging: imaging tools that slot neatly into PACS, ambient documentation saving hours of admin time, clinical trial automation that clears bottlenecks.
These prove one core principle: when AI fits a specific workflow, there is tangible return on investment.
But the bigger opportunity is broader: AI woven through full patient journeys, as well as automating more of the drug discovery and hiring processes. That means fewer missed diagnoses, faster trials, smoother claims and stronger, more resilient teams.
At Emerge, we see this happening through three key channels:
- Workforce development: Training and enabling people to use AI so that less experienced workers become productive faster.
- Knowledge management: Putting the right information in front of people at the right moment so that decisions are faster and safer.
- Workforce enablement: Automating repetitive, low-value processes so that staff time goes further and can be applied to high-value tasks.
Read on for:
- The four market categories where AI is creating the biggest opportunities across care delivery, drug discovery, operations and hiring
- The size of the market ($1.6T spend, ~$490B addressable by AI)
- Which incumbents vs challengers are winning right now – and where founders should build
* * *
Regulators are approving record numbers of AI-enabled devices and AI medical devices, and health systems are already seeing measurable results.
“AI models are now spotting conditions humans were never trained to see. One system can even detect early signs of diabetes from a plain X-ray – patterns clinicians wouldn’t necessarily think to look for.”
Shiv Gaglani, cofounder & CEO, Osmosis and Emerge Venture Partner
From our deep analysis of products that truly break through in this space, we’ve identified three key traits:
1: Fit the work
AI works best when it plugs into the systems staff already use and takes steps away, not adds them. That’s far easier in areas with a single backbone system – like radiology with PACS or pathology with LIMS – than in settings like EHRs, where information is scattered across dozens of platforms. The harder, next evolution will be stitching AI across the full patient journey through longitudinal EHR integration.
Why it matters: Clinicians and ops staff have no slack; anything that adds clicks or complexity will struggle to make it through the pilot. Embedding in existing workflows avoids retraining, speeds deployment and raises switching costs, which increases stickiness and scale potential.
2: Pay back fast
They deliver hard, near‑term wins e.g., revenue uplift per patient, better codification and documentation that increases revenue per case, minutes saved per visit, faster triage and routing, higher theatre/infusion throughput, lower claim denials or quicker trial enrolment. Essentially – ROI that can be felt straight away, giving executives tangible proof points to justify scale and frontline staff immediate incentives to keep using the tool.
Why it matters: Hospital margins are thin and governance is rigorous. Value‑analysis committees and CFOs prioritise in‑year savings/capacity gains; solutions that don’t prove ROI quickly rarely get funded beyond pilots.
3: Compound over time:
They secure consented, ongoing access to operational data so accuracy and automation improve with use.
Why it matters: Data loops create performance advantages: the more data a system sees, the better it gets at its job. That leads to more accurate models, better outcomes for patients and stronger contracts with hospitals. Over time, this turns a small tool that does one job well into a platform that organisations depend on for many jobs.
Caveat: In many markets there is no centralised patient record; interoperability is limited and data sharing often still happens by fax (!). Before advanced data loops can compound value, systems need single sources of truth (structured records, consent flows, interoperability).
… So what?
If a tool doesn’t move a unit‑cost or capacity bottleneck, it stays a point solution. If it does, it becomes part of the operating model and earns expansion budgets.
Breaking down the market
At Emerge, we break down this healthcare market into four core opportunities: Care delivery, Drug discovery & development, Administration & operations, and Hiring & workforce management.
Let’s look closely at each segment – the tools and software currently available, pain points in existing processes, market sizes and openings we are excited to back.
Care delivery
What this includes
- Triage and decision support – spotting problems early and guiding the right next step.
- Documentation and coding – turning the visit into accurate notes and billing data.
- Coordination and handover – making sure information passes safely between teams and shifts).
- Follow‑up – checking in after discharge or appointments so issues don’t get missed.
- Safety nets – systems that catch errors like wrong doses or missed vitals before they cause harm.
How the work currently gets done
- US: Epic Systems (~50% of the market) and Oracle Health are the two main providers of hospital record systems that store patient information, manage treatments and handle billing. Nuance (by Microsoft) supplies speech-to-text tools that help doctors and nurses create notes more quickly.
- Europe: Far more fragmented and country-specific. In Germany alone there are ~10 hospital EHR providers with no single dominant share, and a separate market for ambulatory/patient-management software used by private physician practices. Cross-border vendors are rare, and systems often don’t interoperate. Some hospitals can have as many as seven different healthcare IT systems, each serving a different purpose, such as documentation, bookings, prescriptions, etc.. Epic is expanding their European presence. However, industry insiders claim SAP is pulling back from Europe, creating space for new entrants.
Pain points in the existing process
- Heavy admin load: EHRs were designed to help hospitals get paid and enable public health-level reporting, not to help doctors treat patients. That means doctors have to pick the right billing code from long menus and tick endless boxes just to record something simple. Instead of spending time with patients, they spend hours clicking around confusing screens.
- Poor interoperability: Hospital and ambulatory systems rarely connect; information is still exchanged via phone calls and faxes, adding hours of delay and poor patient experience.
- Decision support: In Europe, many record systems don’t warn enough: for instance, important safety alerts (such as dangerous drug interactions or dose errors) are missing or hard to see. Meanwhile in the US, there are so many pop-ups and warnings, often about low-risk issues, that clinicians start clicking past all of them (“alert fatigue”). Both ends of the spectrum mean real dangers get missed; the fix is fewer, clearer, high-priority alerts that show up at the right moment in the workflow.
- Integration barriers: Major vendors keep critical features in-house; startups must work via hospital IT, making adoption slow and fragmented.
Market size
Existing spend = $60B
Addressable opportunity = $137B
- Reduction in admin, saving ~$200B. Clinicians spend around 4 hours a day on paperwork – roughly 2 billion hours a year across the system – worth about $200B in wages.
- Fewer missed appointments, saving ~$215B. Missed appointments waste capacity (US + UK ≈ $150B, assuming that is 70% of global total = ~$215B globally)
- Medication errors cost $42B globally
If we assume that AI can capture 30% of the value here, $457B * 30% = $137B.
Opportunity spaces where founders should build
Workforce development: Practical training that helps clinicians and support staff use AI safely in the flow of care – e.g., simulation for high‑stress scenarios and communication (Virti; Osso VR), and bitesizes microlearning at the ward level (Elemeno; HealthStream).
Why this matters: Enabling staff to be better prepared means fewer errors, and enabling them to communicate more clearly with patients results in a better level of service.
Knowledge management: Bedside guidance and care guidelines that update in real time, which also surface the right evidence at the right moment – e.g., AI‑augmented reference tools (Amboss, OpenEvidence, Hippocratic AI) and safety‑net alerts/checklists embedded into workflow (MedAware; Lumeon; Aidoc).
Why this matters: Faster, more consistent decisions; fewer omissions. Particularly poignant at the GP stage where symptoms could be any number of issues. Also exciting new data can be analysed using AI.
Workforce enablement: Automation that removes clicks from the clinical day – e.g., ambient notes and coding (Suki), AI imaging triage (Viz.ai, Qure.AI), smart outreach that reduces no‑shows (DrDoctor).
Why this matters: More patient time, shorter waits and fewer missed diagnoses. Furthermore, AI tools are expanding what is diagnosable, e.g. systems can detect early diabetes from x-rays, patterns clinicians are not typically trained for.
“Contrary to the idea that AI will replace people, it’s actually putting them back at the center. In healthcare and beyond, AI tools are cutting paperwork and routine tasks so frontline workers can focus on patient care, judgement, and the parts of the job that truly matter.”
Mehul Patel, CEO @ Ascend Learning, exCOO & President @ Apollo, and Emerge Venture Partner
Administration & operations
What this includes
This covers the administrative and operational backbone of healthcare. It includes:
- Scheduling and patient flow – making sure appointments, beds, and operating theatres are allocated efficiently.
- Claims and revenue cycle management – submitting bills to insurers, correcting errors, chasing denials, and getting hospitals paid.
- Compliance and reporting – producing all the documentation regulators and managers require.
- Data exchange – moving information safely and reliably between different hospital systems, insurers and government databases. Note that given the insurance/cash transaction-heavy nature of US healthcare, this section predominantly pertains to the US market.
How the work currently gets done
Optum provides large-scale claims processing and analytics platforms that help insurers and providers manage billing, reimbursement and population health data. Oracle Health (which now includes Cerner) underpins much of the hospital back-office – from scheduling and patient flow to claims submission and compliance reporting – which essentially constitutes the operational and financial backbone of many health systems.
Pain points in the existing process
- Constant rework on denied claims: A large share of bills bounce back on first submission, forcing staff to fix and resend, clogging up back offices.
- Inefficient scheduling: Beds, appointments and operating rooms are often left unused because planning is manual and forecasting poor.
- Heavy reporting load: Compliance and performance reporting eats thousands of hours, with staff duplicating work across different systems.
- Systems don’t connect: Hospital IT, insurers and government databases rarely talk to each other. Staff re-enter the same data many times, still faxing or emailing files.
- Repetitive admin tasks: Intake, referrals and eligibility checks are still mostly manual, draining time that could be spent with patients.
Market size
Existing spend = $1.4T
- Admin spend = $1.4T. McKinsey estimates $1T is spent on administration for US healthcare annually. Assuming US represents 70% of global total here (as US healthcare is more admin heavy than other geographies), then global market assumed to be $1.4T.
Addressable opportunity = $280B
- A large share of admin spend is avoidable duplication and inefficiency. The biggest pressure points include claims and appeals rework (correcting denied claims), transaction inefficiencies (delays to authorisation), and laborious billing processes (excessive documentation and manuel coding processes).
- Assuming AI and automation can drive a 20% reduction in admin spend (in line with automation benchmarks), this yields $280B annually.
Opportunity spaces where founders should build
Workforce development: Training frontline admin teams to use automation well.
Why this matters: Ensuring human‑in‑the‑loop processes run smoothly through effective training of modern tooling
Knowledge management: Structured rules and playbooks kept up‑to‑date and pushed into workflows.
Why this matters: Fewer errors and escalations; consistent handling of edge cases.
Workforce enablement: Automation tools for admin staff (Notable, LeanTaaS), or tools that enable more streamlined, joined up patient data (OncoFlow).
Why this matters: Better joined up patient data drives better outcomes through more holistic care.
Hiring & workforce management
What this includes
This is all about getting the right staff in the right place at the right time. It covers:
- Recruitment – finding and hiring permanent clinicians and nurses.
- On‑demand staffing – using temporary or travel staff to fill gaps when shortages arise.
- Credentialing and onboarding – verifying qualifications, licences and background checks, and getting new hires up to speed quickly.
- Retention and training – programmes that build skills, resilience and career progression so staff stay and grow rather than leave.
How the work currently gets done
symplr provides credentialing and compliance systems that check clinicians’ qualifications and keep records up to date, ensuring hospitals stay in line with regulations. Alongside this, many hospitals still depend on traditional staffing agencies to fill last‑minute or temporary shifts, often at high cost and with limited flexibility.
Pain points in the existing process
- Not enough applicants for hard‑to‑fill roles and shifts. Shortages in critical units (ICU, theatres, night/weekend cover) mean posts sit open for months.
- Credentialing and checks are slow and fragmented. Verifying licences, right‑to‑work, references and background checks often live in different systems and email threads, pushing start dates back by weeks.
- No single, live view of skills and availability. Managers can’t easily see who is qualified and free, so rostering is done by guesswork and last‑minute phone calls.
- No single source of truth of candidates drives heavy reliance on agencies: Only a third of healthcare practitioners are on LinkedIn, which makes it very difficult for in-house teams to source candidates.
- Clunky onboarding delays productivity. New hires spend days on paperwork and scattered training before they can work at full speed.
Market size
Existing spend = $37B
- Healthcare staffing market valued at $37B.
Addressable opportunity = $28B
- Estimated that using staffing agencies costs $6k per hire, which is 4x that of doing so internally (source: Pickle survey).
- Reducing reliance on agency hiring by empowering internal staffing teams would therefore reduce the $37B spend on agencies by 75% = $28B saving.
Opportunity spaces where founders should build
Workforce development: Programmes focused on onboarding and early‑career readiness for healthcare professionals (OMS, Relias).
Why this matters: Lowers turnover and boosts satisfaction.
Knowledge management: Live, accurate view of supply and skills as well as internal talent marketplaces, and skills graphs that match people to shifts and roles (Compassly).
Why this matters: Ensures you have the right clinician, with the right skills, on the right shift.
Workforce enablement: Tools to enable internal hiring teams to source, screen and schedule clinicians (Pickle, Incredible Health).
Why this matters: Faster, cheaper hiring and better coverage, reduces reliance on agencies saving significant spend.
Drug discovery & development
What this includes
Drug development has several key stages. It begins with target discovery, where researchers identify the biological process or gene to influence. Next comes molecule design, creating or simulating compounds that could work. Then there is trial design and enrolment, setting up studies and recruiting patients. After that, evidence generation through clinical data collection and analysis. Finally, launch and medical affairs, which involves getting regulatory approval, communicating results and monitoring the drug in the real world.
How the work currently gets done
Veeva and Medidata provide the backbone for most drug trials today, covering everything from electronic case report forms and patient data capture to quality monitoring, regulatory reporting and document management. They are deeply embedded; almost every large pharma company and many mid-sized biotechs use them. However, these systems are ageing and haven’t kept pace with modern cloud, AI and workflow design, which creates an opening for newer entrants to replace them with faster, more user-friendly and interoperable platforms.
Pain points in the existing process
- Finding and keeping patients: Clinical trials often fail because not enough people sign up or because participants drop out before the end. This wastes time and money, and can even force trials to shut down.
- Slow trial set-up: Getting approvals and paperwork sorted can take many months. Every delay pushes back the whole development timeline and increases costs.
- Messy data collection: Trial data often sits in different systems that don’t talk to each other. This makes it hard to get a clear picture, slows analysis and increases the risk of errors.
- Over-complicated trial rules: Trial protocols are often so complex, or change so often, that they add months to the process. The extra red tape frustrates teams and can confuse participants.
- Teams working in silos: Scientists, statisticians and regulators often keep information in separate systems and don’t communicate well. As a result, decisions are slower and mistakes slip through.
- Picking the wrong target: Choosing what to study at the start is a big gamble. If the wrong biological target is picked, years of effort and millions in spend can be wasted.
Market size
Existing spend = $126B
Addressable opportunity = $44.8B
- Waste pool = $25B. Global pharma R&D market is ~$247B, and suffers from high late‑stage failure rates of ~90%. Reducing failure rates by just 10% would result in $25B savings.
- Acceleration value = $36B. Drug discovery typically takes 5 years and costs up to $2B; it’s estimated that GenAI could save anywhere from $40-400 million through enhanced data efficiency. With ~90 drug launches in the US/EU per year, this could be anywhere from $3.6-36B. Taking the mid-point of this estimate, we get $19.8B.
Opportunity spaces where founders should build
Workforce development: Training R&D and trial teams to use new data and AI tools confidently – for example, checking model outputs, validating findings and working across departments.
Why it matters: better study design, earlier risk spotting and less wasted effort when trials change direction.
Knowledge management: A single, compliance-ready hub that unifies trial data and documents across pre-clinical, clinical, regulatory and launch stages, so every team is working from the same source of truth.
Why it matters: faster decisions, fewer errors and less time lost hunting through silos of documents.
Workforce enablement: Automation of the most painful parts of trial operations, such as patient recruitment (Autocruitment, Trialbee) and enrolment forecasting, and report drafting (Medable).
Why it matters: shorter delays, fewer failed studies and a quicker path to approval for critical drugs.
“The greatest risk to a billion‑dollar drug launch isn’t the science; it’s the fracturing of core knowledge and evidence across siloed teams. AI can bridge these silos, reduce time to information and empower launch teams to act with the speed, clarity and compliance that the market demands.”
Ragnor Comerford, founder, EQTR
Conclusion
What feels like science fiction today will become infrastructure tomorrow.
The winners won’t digitise old processes; they will re-architect how work gets done. They will train people faster, surface knowledge at the point of care, turn wasted hours into patient time and transform fractured data into decisions.
“Medical knowledge is doubling frighteningly fast. A GP can face hundreds of conditions and thousand-page guidelines that change every few years. AI that filters what’s relevant is no longer a nice-to-have”
Sievert Weiss, cofounder, AMBOSS and Emerge Venture Partner
