Details
About the Reviewer
Reviewer Organization
Reviewer Tech Stack
Other Products Considered
Summary
Product Usage: The Nabla Copilot AI software is used to provide real-time documentation during patient visits, generating transcripts and summaries within seconds.
Strengths: The software excels in providing quick, reliable transcription services, with a user-friendly and efficient interface.
Weaknesses: It can struggle at times with shorter patient interactions due to insufficient data and lacks personalization in its note-taking format.
Overall Judgment: While there are a few areas needing improvement, the reviewer believes Nabla Copilot was a good choice, lauding its ease of use, installation, and reliability.
Review
So today we’re chatting about Nabla Copilot and how it’s used at your company. Before we jump into that, could you give a brief overview of the company and your role there?
We are a digital health provider specializing in primary and urgent care. I’m a physician and the practice owner.
What problem were you looking to solve in searching for a tool like Nabla?
We aimed to streamline information capture during patient visits. Documenting post visit increases workload, whereas real-time documentation can detract from the patient-physician interaction, especially in a video consultation setting. We needed a tool that would allow us to provide our full attention to the patient while minimizing post-visit documentation.
Did you have any specific requirements that you were looking to fulfill?
The key feature we needed was ambient AI capability. The tool had to function effectively in a remote setting, not just in-office. It was essential that the system operate quickly, providing real-time documentation without significant delays—we couldn’t wait 5 or 10 minutes for a note after the visit. Additionally, the accuracy of the tool needed to be high.
How were you solving this problem prior to Nabla?
In previous roles, I used Dragon for transcription purposes. Although it’s meant for dictation rather than scribing, I found that as a fast typist, the benefits were minimal. The software occasionally misinterpreted words, which led me to conclude that it didn’t meet my needs. I was searching for a tool that could do more than simple dictation; I needed something that could summarize patient encounters effectively as they happened in real-time.
Did you look at alternatives to Nabla, and if so, how did they compare?
I looked at Suki, which didn’t offer ambient AI at the time, so that was a limitation. Nabla stood out due to its simplicity in setup and use, which was a major advantage. I also looked at Abridge, but it seemed more suited for larger customers, lacking a trial option and requiring interaction with a sales representative, which indicated a focus on enterprise-level sales. This approach wasn’t ideal for smaller practices, and at the time, seven months ago, there were no other competitors that fit the bill for a quick and easy trial targeted at smaller practices.
How was the sales and onboarding process?
The process is completely self-service and very efficient. You install a Chrome extension, or alternatively, there’s also an app available. Upon signing up, they offer a certain number of free patient interactions each month, allowing for immediate trial use. After initiating the service, a representative from Nabla contacted us to extend a welcome and discuss further use. However, no direct communication was necessary to begin using the platform; it was set up within minutes.
How does Nabla fit into your workflow?
In virtual encounters, the patient and physician talk, and Nabla generates a note at the end of each visit, typically within 5-10 seconds. I usually have to make some adjustments to the notes, and while they aren’t perfect, Nabla is continually working to enhance their accuracy. Another use case for us is the real-time transcript. If I miss something a patient says, I can quickly refer back to the transcript without interrupting them. This assists in maintaining a seamless conversation, and although it’s not Nabla’s primary feature, it’s a valuable addition.
What types of revisions have you had to make to the notes, and is there any general pattern that you’re seeing in terms of what needs to be amended?
Things like correcting medication names that are sometimes inaccurately transcribed. I also remove information deemed not pertinent to streamline the notes, but that’s something that varies according to physician preference. I aim for conciseness and relevance.
In large language models, I’ve observed changes over time in their output, possibly due to feedback and behind-the-scenes adjustments. For instance, Nabla used to always provide an assessment and plan, but recently, rather than entering something, it has started indicating that there’s insufficient information to generate these sections. This new behavior is a noticeable shift from its earlier versions and can be frustrating, since the input hasn’t changed to necessitate such a response. It may be an attempt to ensure accuracy, but it requires additional manual input, which is needed in perhaps 5% to 10% of encounters, especially with simpler patient cases.
What types of appointments do you have, and how long do they usually last?
Our practice includes primary care and urgent care type visits, where visit lengths vary significantly. Some are brief, like a quick request for a refill on a medication, while others are more complex, lasting up to 20 minutes. In contrast, primary care intakes are more uniform in duration.
How well do you find that Nabla performs across those various types of visits?
The system performs better with complex visits, as it has more data to analyze. In contrast, with shorter visits, it sometimes cannot generate an assessment and plan due to insufficient information.
How do you feel about the actual product experience, the UI?
The software is user-friendly, quick, and ensures that notes are never lost, even if Chrome crashes or the browser is accidentally closed. It offers various templates, including classic SOAP notes and specialty-specific notes like psychiatry, cardiology, and lactation. However, customization based on individual provider preferences could be better, such as allowing the system to learn and adapt from corrected notes, although privacy concerns may limit this capability. The company has thoroughly communicated their commitment to patient privacy on their website, detailing the technical measures they take to protect this information. But it would be beneficial if the software could learn from individual providers and tailor notes accordingly while maintaining privacy standards.
Do you typically edit your note within the Nabla UI before copying it, or do you first copy it over and then edit it?
You can dictate additions to a note, and they’ve introduced dot phrases, which are essentially templates. For example, for a COVID case, you can insert a pre-made COVID text. While these features allow you to add content, I don’t think you can edit what’s been created. So I tend to copy the entire note, paste it elsewhere, and then make my edits.
What would you characterize as the relative strengths and weaknesses of Nabla?
The strengths are its ease of use and installation, along with its reliability and the fact that you don’t lose notes. The weaknesses include the system’s occasional difficulty in generating assessments and plans for simpler patient encounters, although it performs well 90% of the time, particularly with more complex patients. Another limitation is the lack of customization options to tailor the output to specific user preferences.
Comparatively, I’ve experimented with Chat GPT by simulating a patient visit and found it capable of generating SOAP notes on command. It can even craft a differential diagnosis within the assessment and plan if prompted correctly. It would be beneficial if Nabla could incorporate similar functionalities that allow for such prompt-based customizations.
How reliable and bug-free is the platform itself?
It’s always been reliable and available whenever I’ve needed it. There was a bug early on, but I don’t recall the specifics. Since then, I’ve never encountered any issues, and there haven’t been any recent bugs.
How would you characterize the support you’ve received?
I had an issue at first; it was a minor bug, possibly a spelling error or a grammar issue. I mentioned it, and the support team responded immediately. That was my only interaction with support.
Looking back over the several months that you’ve been using it, do you feel like you made the right decision in going with Nabla?
I’m satisfied with the product and feel that I made the right decision at the time. However, I’m not fully aware of the current alternatives in the market.
Do you have any general advice for buyers who might be in a similar space?
When selecting a tool, it’s important to carefully consider your choices due to the inertia that comes after implementation. I suggest taking the time to explore options and read reviews. For example, I simply searched for AI scribes, but now we have lists like this one from Elion that incorporates user feedback and trial experiences. Even with user-friendly options like Nabla, which requires minimal setup and technical investment, there’s still a tendency to stick with what you’re used to. However, there might be better tools available. Nabla is especially easy to try, which might not be the case for all tools. If Nabla meets your needs quickly and effortlessly, its convenience might be a deciding factor.