From AI scribe to AI care partner: Tandem Health is moving into CDSS and aftercare

Dr Kola Tytler (MBBS CertHE MBA MRCGP)||13 min read
Featured image for From AI scribe to AI care partner: Tandem Health is moving into CDSS and aftercare

AI scribes used to promise one thing above all else: they would write the note so the clinician could focus on the patient.

That promise still matters. It is also no longer the whole story.

The latest public signals from Tandem Health suggest something more ambitious is taking shape. Tandem is no longer positioning itself only as a documentation tool. It is describing its ambition as building an AI-native operating system for clinical workflows, and it has publicly said that this next phase includes:

  • capturing the right clinical codes
  • coordinating care
  • supporting decisions at the point of need

At the same time, its product surface now spans much more than ambient documentation:

  • notes and document generation
  • coding support
  • referral and letter generation
  • “Secretary” workflow for handover, referrals, letters, scheduling, and follow-ups
  • one-click EHR transfer
  • positioning as an AI assistant for every step of care

That combination matters.

It suggests a category shift from AI scribe to AI care partner — and, potentially, toward an increasingly decision-adjacent platform that sits before, during, and after the consultation.

This article unpacks what Tandem has publicly said, why CDSS is a much bigger step than documentation, why aftercare is the natural next workflow layer, what the risks of “everything platform” thinking are, and where specialist tools still win.

The wedge is simple:

Documentation support and aftercare orchestration do not replace guideline-first pathway execution.

The short version

If you want the whole article in one practical summary:

  • Tandem began with documentation relief.

  • It is now clearly signalling a move toward a broader clinical operating layer.

  • Publicly, that ambition includes decision support at the point of need, coding, and care coordination.

  • Its Secretary product and document-generation workflow make aftercare a very obvious next frontier.

  • This is strategically powerful because the note is not the endpoint of care — it is the start of the next set of actions.

  • But there is still a major distinction between:

    • capturing and moving clinical work forward, and
    • guiding the clinician through thresholds, escalation logic, and pathway execution.

That distinction is exactly where specialist, guideline-first tools still matter.

What Tandem has publicly said

The strongest basis for this article is Tandem’s own language.

In its Series A announcement, Tandem said it is building an “AI-native operating system for clinical workflows across Europe” and explicitly stated that the next phase includes:

  • capturing the right clinical codes
  • coordinating care
  • supporting decisions at the point of need

That is not routine “scribe feature expansion” language. It is operating-system language.

Separately, Tandem’s public product pages now frame the product as an AI assistant for every step of care, and explicitly say Tandem helps clinicians prepare, document, and follow up on visits.

Its released products and product surfaces now include:

  • clinical notes
  • document generation
  • referrals, sick notes, patient summaries, and letters
  • coding suggestions
  • EHR transfer
  • and Tandem Secretary, which is explicitly about referrals, letters, scheduling, and follow-ups inside a shared workflow between doctors and secretaries

Taken together, that is enough to support a careful conclusion:

Tandem is moving beyond note generation into a broader care-workflow and decision-adjacent platform.

That is the real story.

Why this is more than “they added features”

When AI companies add features, it is tempting to treat each one as incremental:

  • now they do notes
  • now they do letters
  • now they do coding
  • now they do referrals

But this is not just incremental expansion.

It is a change in what part of clinical work the company is trying to own.

A pure scribe lives mostly in one moment:

  • the consultation note

A clinical workflow operating layer lives across multiple moments:

  • before the consultation (preparation)
  • during the consultation (capture / note support)
  • immediately after (coding, documents, referrals, patient summaries)
  • the aftercare layer (follow-up tasks, handover, status updates, coordination)
  • and, increasingly, decision-adjacent prompts and support at the point of need

That is a completely different category ambition.

Why CDSS is a step-change compared with documentation

This is the most important part of the whole story.

Documentation support and CDSS are not the same thing.

What documentation support does

Documentation support mainly helps with:

  • reducing admin burden
  • capturing the consultation more efficiently
  • improving presence with the patient
  • speeding up note and letter production

This is already useful, and there is strong real-world appeal in it.

What CDSS does

Clinical decision support starts touching a different layer of work:

  • what to consider next
  • what threshold changes management
  • what code fits the encounter
  • what route is appropriate
  • what additional action, referral, or follow-up may be needed
  • what needs checking at the point of need

That is a much more sensitive zone.

Why?

Because once a system starts shaping the clinician’s interpretation of the case — or influencing what happens next — it is no longer just reducing typing. It is entering the space of clinical judgement support.

That is where the stakes rise sharply.

Why Tandem’s public wording matters

Tandem has already said the operating-system ambition includes supporting decisions at the point of need. That is exactly the kind of phrase clinicians and buyers should pay attention to.

It does not automatically mean “AI is making the diagnosis”.

But it does mean the product direction is moving from:

  • recording what happened

toward:

  • helping shape what happens next

That is a major category shift.

Why aftercare is the natural next layer

If documentation is the first beachhead, aftercare is the obvious adjacent market.

Why?

Because the note is rarely the end of the work. After the consultation, someone still needs to:

  • send the referral
  • generate the letter
  • arrange follow-up
  • task the admin team
  • chase results
  • send patient information
  • code the encounter correctly
  • make sure nothing gets dropped between the note and the next action

This is where the work often becomes fragmented.

Email chains, EHR task lists, letters, secretarial handoffs, referral systems, and follow-up admin can all pull clinicians and teams out of flow.

That is exactly why Tandem Secretary is such an important signal.

It is explicitly designed to connect doctors and secretaries in a shared workflow for:

  • referrals
  • letters
  • scheduling
  • follow-ups
  • status updates tied to the source note

In other words, Tandem is not just trying to write the note faster. It is trying to own more of the post-consultation handover layer.

That is aftercare infrastructure.

Why the aftercare layer is strategically powerful

From a product and market perspective, aftercare is one of the most attractive workflow layers to own.

1) It is repetitive and expensive

Referrals, follow-up admin, coding, and coordination consume huge amounts of time.

2) It sits right after the note

If you already generated the note, you already hold much of the structured information needed to create:

  • referral letters
  • patient summaries
  • sick notes
  • follow-up tasks
  • coding suggestions

That makes aftercare a natural extension of the scribe.

3) It increases workflow gravity

A documentation tool can be turned off. A workflow layer that also handles handovers, tasks, and post-visit coordination becomes much harder to replace.

4) It changes the buying logic

Once the product covers:

  • note capture
  • document generation
  • coding
  • handoff
  • follow-up workflows

…it starts to look less like “an AI feature” and more like workflow infrastructure.

That is exactly why the operating-system framing matters.

Why the UK/NHS angle makes this even more interesting

This is not just a European private-market story. There is a real NHS angle.

Accurx has publicly said that Accurx Scribe, powered by Tandem, is being rolled out across its NHS footprint and can generate:

  • consultation summaries
  • documents
  • referral letters
  • Advice & Guidance requests

That matters because once a scribe starts generating A&G requests and referral letters, it is no longer just “documenting the encounter”. It is already touching the handoff into the next stage of care.

Separately, Tandem appears on NHS England’s Ambient Voice Technology Supplier Registry, which places it inside the broader NHS adoption and assurance conversation.

This means the story is not merely “interesting startup strategy”. It has direct implications for UK clinicians and for how AI infrastructure may spread through NHS workflows.

The hidden risk: when “care partner” becomes too vague

This is where some scepticism is healthy.

The phrase AI care partner can sound exciting, but it can also blur important distinctions.

Because not all “care partner” functions are equal.

There is a big difference between:

  • generating a referral letter
  • handing a task to a secretary
  • suggesting a clinical code
  • surfacing next-step information
  • and guiding real clinical decision-making under uncertainty

If those activities get bundled under one broad “care partner” story, there is a risk that clinicians and buyers stop asking the right question:

Exactly which jobs is the system helping with, and how close is it getting to clinical judgement?

That question should never be skipped.

The risks of “everything platform” thinking

The move from scribe to operating layer is understandable. It is also where new risks emerge.

1) Workflow convenience can hide decision risk

A tool that saves clicks and keeps everything in one place will often be adopted more readily — even if it is not actually the best tool for every task.

Convenience is powerful. But convenience is not the same as correctness.

2) Documentation confidence can spill over into decision confidence

If clinicians come to trust a platform because it writes good notes, they may begin to trust it in adjacent decision-support tasks more than is warranted.

That is a subtle but important risk.

3) Bundling can blur which layer is responsible for what

Once notes, codes, referrals, and decision-adjacent prompts live in one product, it becomes harder to separate:

  • what is administrative support
  • what is workflow coordination
  • what is clinical decision support
  • what must still be independently verified

4) Aftercare automation can create a false sense of completion

A task being generated, delegated, or marked complete is not the same thing as the right pathway having been followed.

This matters especially in threshold-based, guideline-heavy systems like UK primary care.

Where specialist tools still win

This is the most important balancing point.

Tandem’s direction can be strategically smart and still leave major gaps that specialist products fill better.

The strongest specialist moat remains this:

Documentation support and aftercare orchestration do not replace guideline-first pathway execution.

What specialist tools still do better

A specialist knowledge product can still outperform a broad workflow platform when the clinician needs:

  • clear thresholds
  • structured escalation logic
  • practical what-next pathway support
  • rapid review of a condition-specific workflow
  • a more explicit bridge between national guidance and clinical action
  • structured reasoning support when the case is messy

This is particularly true in UK-style primary care workflows, where the real bottleneck is often not “write the note” but:

  • “Do I need A&G or direct referral?”
  • “What is the threshold here?”
  • “What needs doing before referral?”
  • “Is this red-flag enough to escalate now?”
  • “What does the guideline pathway actually expect next?”

Those are not purely documentation problems. They are pathway problems.

Where iatroX fits in this new stack

This is where iatroX can position cleanly and credibly.

The point is not to compete with Tandem on note-writing or workflow orchestration. The stronger position is to become the knowledge hub at the point of need inside a broader clinician AI stack.

Use iatroX when the job is:

  • pathway review
  • thresholds and escalation logic
  • structured clinical Q&A
  • rapid guideline refresh
  • messy-case reasoning before or after the consultation

Core iatroX routes to link in this article

The clean positioning line

Tandem may increasingly help clinicians capture, code, delegate, and follow up.

iatroX can still be the place clinicians go when they need to decide, clarify, and execute the pathway safely.

That is not overlap. It is stack design.

A practical way clinicians should think about this category shift

If you are a clinician or buyer, the most useful framing is not:

“Will AI scribes replace everything else?”

It is:

“Which parts of the workflow should be bundled, and which parts should remain specialist?”

A sensible answer might look like this:

Bundle into the workflow platform

  • note capture
  • document generation
  • coding
  • task handoff
  • referrals and letters
  • follow-up coordination

Keep specialist

  • pathway execution
  • threshold-based decision support
  • structured reasoning
  • guideline-first review
  • learning and retention

That is probably where the market is heading. Not one tool for everything — but a more coherent stack with clearer roles.

What this means for the broader AI scribe market

Tandem is not the only company moving this way.

But its public language is useful because it makes the shift explicit:

  • not just an AI scribe
  • not just a document assistant
  • not just coding
  • but an AI-native operating system for clinical workflows

That framing matters because it tells you what the next battle is really about.

The next battle is not who writes the note best.

It is who owns the workflow after the note — and whether that layer expands into decision support, care coordination, and follow-up infrastructure.

That is a much larger ambition.

FAQ

Is Tandem still “just an AI scribe”?

No. Based on Tandem’s own public positioning, it is moving well beyond documentation into coding, care coordination, and decision-adjacent support at the point of need.

Is Tandem openly saying it is building CDSS?

Tandem’s public language is careful, but it has explicitly said it wants to support decisions at the point of need as part of its AI-native operating system for clinical workflows. That places it in a decision-adjacent direction, even if clinicians and buyers should still ask for precision about what that means in practice.

Why is aftercare such a big deal?

Because the note is not the end of the consultation workflow. Referrals, follow-ups, tasks, letters, coding, and handover consume major time and create fragmentation. Owning that layer is strategically powerful.

Does this make specialist tools less relevant?

No. It makes specialist tools more important where the job is not workflow orchestration but pathway clarity, thresholds, escalation logic, and structured clinical reasoning.

Bottom line

The cleanest way to read Tandem Health’s latest direction is this:

It is moving from AI scribe toward AI care partner.

That shift is visible in:

  • operating-system language
  • explicit references to point-of-need decision support
  • coding as a regulated infrastructure layer
  • Secretary/follow-up tooling
  • broader product framing around preparing, documenting, and following up on visits

This is strategically significant because it means the market is no longer only about removing documentation burden. It is increasingly about shaping what happens next.

But the most important distinction still stands:

Documentation support and aftercare orchestration do not replace guideline-first pathway execution.

That is where specialist knowledge tools still win.

And that is why the most sensible future for clinicians is not a single AI tool that “does everything”, but a stack in which workflow platforms handle capture and coordination, while specialist tools like iatroX remain the point-of-need knowledge layer for thresholds, pathways, escalation logic, and structured reasoning.

Suggested source links

Share this insight