Ambient AI + Dictation: The Full Clinical Day

Clinical care doesn’t happen in neat categories.

It happens in fragments: during the encounter, between patients, early in the morning, late at night. A note that needs one more section. A referral that has to go out. An inbox message that can’t wait. Over time, those moments have been split across different tools, each with its own workflow, contract, and cost.

Ambient AI has changed part of that picture. But only part.

We spoke with Nabla Chief Product Officer Laurent Landowski and Chief Clinical Product Officer Matt Sakumoto about what was still missing, and why dictation was the natural next step toward something simpler: one system for the entire clinical day.

Nabla is known for ambient AI. Why build dictation at all?

Laurent: Ambient only solves part of the problem. It handles the encounter, the conversation between a clinician and a patient. But the full documentation day is much more broad. Physicians are dictating inbox messages early in the morning, correcting notes after complex visits, drafting referrals between patients.

All of that work was still happening somewhere else: in a separate tool, under a separate contract, with a separate workflow. Once we saw that, it didn’t feel like an expansion. It felt inevitable.

Matt: From the clinical side, it was one of the most consistent pieces of feedback we heard. Clinicians would tell us how much they valued the ambient experience, and then immediately point out that they were still switching tools for everything else.

It created another split in their day with modern AI in one moment, and a Dictaphone from the 1970s in another. That kind of fragmentation adds up. It wasn’t how clinicians think about their work, and it wasn’t how we wanted the product to evolve.

What were clinicians actually asking for?

Matt: The requests were pretty grounded. First, they wanted dictation to work directly inside the EHR—at the cursor, where they’re already documenting. Then, they wanted their existing workflows to carry over. Things like dot phrases, template triggers, and voice commands. Those aren’t small conveniences. Muscle memory and micro-efficiencies, multiplied by hundreds of patients a day, have a significant impact. Asking clinicians to rebuild that from scratch isn’t realistic.

Laurent: And in the end, there was one simple ask: don’t make me think about which tool to use. Clinicians are already making constant decisions throughout the day. Choosing between different documentation tools shouldn’t be one of them.

How does dictation fit with ambient AI? Are these competing approaches?

Laurent: They’re complementary. Ambient AI is passive, it runs in the background during the encounter and generates a structured note. Dictation is active, it’s for when a clinician knows exactly what they want to say and needs to say it quickly.

These are different modes, but they belong in the same place. The goal isn’t to choose between them. It’s to have both available, without friction.

Matt: Clinically, it’s pretty straightforward. Ambient handles the encounter. Dictation handles everything around it. If you only have one, there’s always a gap, and clinicians will find another tool to fill it.

Bringing both together closes that loop.

Health systems are under pressure to simplify their tech stack. How do you think about that?

Laurent: Most health systems didn’t intentionally build a fragmented documentation stack. It happened over time and new tools were added as needs evolved.

Now they’re managing multiple vendors, contracts, and integrations. That complexity shows up in cost, but also in operational overhead and clinical friction.

There’s a better path. Not by adding another tool, but by consolidating around an AI solution that already fits into the clinical workflow.

Matt: And consolidation only works if clinicians actually use the product. That’s where familiarity matters.

We were very deliberate about making dictation feel natural from day one, supporting existing commands, templates, and hardware. We already launched at one large health system, and clinicians logged nearly 200 hours of dictation in the first 30 days, without a formal training program. That kind of adoption tells you the transition isn’t disruptive.

What does it mean to have both ambient and dictation in one place?

Matt: For clinicians, it removes friction. The documentation day has a lot of different moments—in the room, between patients, after hours catching up on notes.

When the same system supports all of those moments, it stops feeling like separate tasks. It just becomes part of how you work.

Laurent: For health systems, it simplifies ownership. One integration, one support relationship, one roadmap.

That matters—not just financially, but operationally. The goal is to reduce the number of moving parts, not increase them.

Can you describe what this looks like in practice for a clinician?

Matt: A hospitalist gave me a great example recently. She’s rounding in the morning, moving quickly across patients. Ambient is handling the encounter documentation, so she’s not reconstructing notes afterward.

But then the rest of the day kicks in. She needs to respond to a message, update her sign-out, and dictate a discharge summary. Previously, that meant switching tools and breaking her flow.

Now it’s the same experience throughout. Same interface, same commands, same mental model. She told me it was the first time her day felt like one continuous workflow instead of a series of interruptions. That’s exactly what we’re aiming for.

In my primary care practice, I see the same pattern. About 80% of the work happens outside the exam room, including care coordination, patient portal messaging, and staff communication. That’s where dictation really matters.

Where does dictation go from here?

Laurent: We see this as a foundation. The combination of clinical-grade speech recognition and deep EHR integration opens up a lot of possibilities.

What we’re focused on next is context—understanding not just what a clinician is saying, but what they’re trying to accomplish, and helping them get there more efficiently.

There’s still a meaningful gap between documentation burden and what technology can relieve. We’re working to close that over time.

Matt: And from a clinical perspective, different specialties have very different needs. Areas like surgery, radiology, and pathology rely heavily on precise, structured dictation.

Getting that right requires a deeper understanding of those workflows. It’s something we’re investing in deliberately.

Final question: if you had to summarize this in one idea, what is it?

Laurent: Clinicians don’t experience documentation as separate categories. It’s one continuous responsibility.

Matt: And the tools should reflect that. One system that supports the entire day—not just part of it.

‍