Deep dive into Doximity

Written in partnership with Doximity

I approached this deep dive by looking at how Doximity’s tools behave in real clinical scenarios - testing them with hypothetical cases drawn from A&E, General Surgery, Urology, and Acute Medicine, the areas I locum in, with a particular focus on DoxGPT and Scribe. The aim was to observe where they’re useful, where they fall short, and how they may evolve. To make this evaluation possible, Doximity granted me full access to their platform.

Founded in 2010, Doximity began as a professional network for US clinicians. Over time, it expanded beyond networking to include workflow tools designed to support communication in clinical practice, such as secure messaging, fax and phone calls via Dialer. These tools were added incrementally, building on an existing clinician user base rather than replacing it. Today, over 80% of US physicians are registered members of the platform.  

Recently, Doximity added two new AI products: DoxGPT and Doximity Scribe. These tools can be used on both desktop and mobile, although the mobile app is not available in the UK App Store. Insights shared in this article are based on the desktop version.

Doximity’s expansion into AI builds on its existing platform rather than sitting alongside it. The development of DoxGPT reflects a broader move to integrate clinical decision support directly into tools clinicians already use. This direction was reinforced by the acquisition of Pathway AI in 2025, which brought structured, guideline-based clinical reasoning into Doximity’s AI efforts.

DoxGPT is their AI clinical decision support tool. It acts as an assistant for clinicians, designed to quickly help with administrative tasks and answer clinical questions with peer-reviewed references from over 700 medical journals. 

I explored how the tool handled a pattern-recognition challenge. An ED colleague once encountered a patient with leg weakness associated with chest pain, and the presentation initially caused uncertainty. A senior attending immediately recognised the possibility of an aortic dissection, and the patient received timely imaging. So I wanted to see how DoxGPT handles this scenario. 

• Latency: Answers arrived in under two seconds - the fastest of any AI-CDS I have used so far. Other tools can take significantly longer, and when you’re working a Saturday night ED shift with patients in the corridor, that difference matters.

• Format: Responses are not blocks of paragraphs but structured lists, bullet points and tables. The layout is intentionally easy to skim, making it possible to extract the key actions quickly.

• Traceability: Every answer links directly to a peer-reviewed source, and many include the exact quoted section (1) of the paper, accessible with a single click. This level of transparency is important for clinical use and is something many tools still lack in my experience.

• Access to paywalled papers: The tool also offers access to paywalled papers cited in its responses. A full-text request can be made with one click (2), and users receive five paper credits per month.

• Feedback loop: A feedback mechanism is built into the interface. Many AI tools do not offer this, leaving you unsure how your feedback informs the product. Feedback goes directly to the product team and is monitored closely - an important part of improving systems used by clinicians and ensuring they are heard. 

One area where clinicians often lose time is medication research - checking dosing, contraindications, adjustments or looking up unfamiliar treatments. A significant proportion of advice sought from haematology or microbiology could be avoided with quicker access to reliable information.

Doximity has addressed this thoughtfully. When I typed an antibiotic-dosing query, the answer appeared instantly - even before I pressed enter - and importantly, no AI was used. The tool displayed the exact reference from the FDA label. This reduces uncertainty in high-stakes decisions and offers a more direct route to trustworthy information, without requiring clinicians to navigate full guidelines during busy shifts.

Clicking Drug Monograph opens a clear, comprehensive overview of the medication without leaving the platform.

Beyond clinical decision support, DoxGPT can:

Each new clinical conversation with DoxGPT earns 0.5 CME points. You are essentially progressing your professional development while using the tool.

Doximity’s AI Scribe supports:

It also integrates with Doximity’s dialer and video calls, meaning phone and video visits are transcribed automatically.

Clinical use & the future of workforce

AI-CDS systems like DoxGPT give clinicians a way to enhance their practice through augmentation. One area stands out in particular: when clinicians need rapid answers instead of waiting for specialty or senior input and time does not permit, which is the reality of an overstretched and understaffed workforce today. As healthcare moves toward more distributed and supported decision-making, these systems will become a core part of modern clinical workflows, and Doximity is already offering an early example of what that shift may look like.

DoxGPT is currently available to doctors, NPs, PAs, pharmacists, and students. As adoption grows, policymakers will need to consider how systems like this fit across different points of the care pathway - not only for individual clinicians, but within the wider organisation of the workforce. With physician shortages ongoing, these tools are unlikely to remain limited to doctors. Allied health professionals may take on larger clinical roles supported by AI, provided the right safeguards are in place.

Strategic positioning 

Doximity’s breadth - AI-CDS, scribe, telehealth and fax - is tightly integrated across the platform, with DoxGPT and the scribe offered free to members. This suggests that Doximity is intentionally positioning these AI tools on a distinct axis within its broader ecosystem. At present, neither DoxGPT nor the scribe integrate directly with EHRs, whilst Doximity Dialer does integrate with Epic Haiku, allowing telehealth encounters to be recorded directly in the patient’s chart. While other AI healthcare tools are competing for EHR integrations, my sense is that Doximity is taking a different approach. With its existing market saturation, it may not feel the same pressure to integrate and may prefer to compete on a separate axis.

EHRs can be unwieldy, and operating outside them does offer speed and flexibility. The trade-off is that an AI-CDS that can read the chart is better placed to surface relevant answers when something has been overlooked or the wrong question is prompted. Similarly, the scribe would be more useful if it could summarise previous events or support pre-charting. My assumption is that Doximity is aware of these gaps and may already be planning how to address them.

Whether this standalone approach becomes an advantage or a limitation will depend on how clinical workflows evolve, but Doximity’s foundation and scale place it in a strong position to help shape whatever that future looks like.

*_{This deep dive is part of a sponsored collaboration with Doximity, though the analysis is entirely my own. I retained full editorial control throughout, and the company was open to a neutral, balanced evaluation - including limitations. I regularly analyse health AI tools, and this deep dive follows the same structure and standards.}