Executive summary
The clinical decision support system (CDSS) landscape is undergoing a rapid and transformative evolution. In 2025, the conversation is dominated by Retrieval-Augmented Generation (RAG) technology, conversational search, and the promise of seamless integration into ambient clinical workflows. This shift is underscored by significant market signals, including Doximity’s strategic acquisition of Pathway to bolster its AI suite, the rapid scaling of new entrants like Glass Health, and the continued innovation from major incumbents such as UpToDate AI Labs and EBSCO’s Dyna AI (Doximity Investors, Yahoo Finance, Fierce Healthcare, Wolters Kluwer, EBSCO).
For UK clinicians, the promise of speed and intelligence must be balanced with the professional necessity for transparent sourcing, regulatory alignment, and practical workflow integration. As NHS England’s guidance on implementing clinical decision support emphasizes, the true measure of these tools lies in their ability to deliver safe, evidence-based improvements in patient care (NHS England). This article maps the key players and trends shaping the future of AI-powered clinical reference.
Why this market is heating up
Three key forces are driving the intense momentum in the AI-powered CDSS market:
- Consolidation and capability acquisition: In a landmark deal, US-based physician network Doximity acquired Pathway, a Montreal-based AI clinical reference platform. The move brings Pathway's "physician AI team + datasets" into Doximity's growing suite of tools, which already includes Doximity GPT and a free AI scribe (Business Wire, Doximity Investors).
- Category validation by incumbents: The market leaders have fully embraced generative AI. UpToDate’s AI Labs and Enterprise Edition, along with EBSCO’s Dyna AI, are pushing the standard for grounded, citation-first answers delivered directly at the point of care, validating RAG as the core technology for trustworthy clinical AI (Wolters Kluwer, EBSCO).
- New entrants scaling fast: Innovative startups are attracting significant attention and funding. Glass Health, a seed-backed company, is gaining traction with its focus on using generative AI to help clinicians build differential diagnoses and draft clinical plans, directly targeting the problem of physician burnout (blog.glass.health, wilmerhale.com).
Company spotlights (what they actually do today)
Doximity + Pathway
Pathway’s core product offers clinicians concise guideline summaries, interactive algorithms, and differential diagnosis support. Recent features include automated note generation, educational quizzes, and CME-earning Q&A modules (pathway.md). The acquisition, with a headline value of $63 million, is set to accelerate Doximity’s AI roadmap, creating a potential one-stop hub for clinicians that combines a newsfeed, telehealth capabilities, an AI scribe, and now, a powerful clinical reference tool (Yahoo Finance, doximity.com, Fierce Healthcare).
Glass Health
Glass Health has positioned itself as an AI-native CDSS designed specifically to help with the core reasoning tasks of forming a differential diagnosis (DDx) and drafting a management plan (Glass Health, Y Combinator). As a seed-backed startup, its product narrative is tightly focused on leveraging generative AI to augment the clinician's thought process and reduce the cognitive load associated with complex cases (blog.glass.health).
Established CDSS baselines clinicians already know
- UpToDate (Wolters Kluwer): The global reference standard continues to innovate with its AI Labs and an Enterprise Edition that features AI-enhanced search and user analytics, with more advanced GenAI features in ongoing testing (Wolters Kluwer).
- DynaMedex / Dyna AI (EBSCO): This platform is a strong proponent of RAG-grounded answers with highly transparent sourcing. Its Dyna AI feature is now available to individual subscribers, bringing its powerful, evidence-based Q&A capabilities to a wider audience (EBSCO).
iatroX (disclosure)
iatroX provides UK clinicians with evidence-linked Q&A and structured brainstorming tools for educational and reference purposes, grounded exclusively in national guidelines and peer-reviewed research. It is designed to provide rapid, reliable information to support, not replace, clinical judgment. Our team is actively exploring partnerships with complementary platforms to widen trusted content coverage and deepen workflow integrations.
How these tools differ (and why it matters at the bedside)
- Knowledge provenance: A key differentiator is how tools generate answers. The transparent, RAG-grounded approach of Dyna AI and UpToDate AI Labs, which provides native links to primary literature, contrasts with more summary-first experiences.
- Scope: The tools are increasingly specialised. Glass Health focuses on DDx and plan generation; Pathway excels at guideline look-ups and interactive algorithms; UpToDate remains the encyclopaedic reference with graded recommendations; and DynaMedex offers exceptional depth in drug monographs and medicine-focused queries.
- Workflow bundling: The strategic direction of Doximity—bundling a scribe, reference tool, and messaging in one platform—could be a powerful way to reduce context switching, provided the execution preserves the speed and citation quality clinicians demand (Fierce Healthcare).
Buyer’s matrix (what a GP/registrar, hospitalist, or pharmacist should evaluate)
- Evidence & transparency: Are answers linked to primary sources and/or guideline IDs by default?
- Regulatory posture: Does the vendor have a clear statement on its medical-device status? Does it align with NHS CDS implementation guidance?
- Integration: Is there API availability or SMART on FHIR support for EHR embedding? Are single-sign-on (SSO) and audit logs available?
- Coverage & speed: Does the tool have the necessary specialty depth, algorithm breadth (Pathway), or DDx quality (Glass Health) for your needs?
- Commercials & risks: Understand the total cost of ownership (licences plus IT support), how patient data is handled, and the vendor's post-acquisition roadmap credibility.
Practical use cases to include (role-based)
- GP / hospitalist: Perform a "30-second check" of management steps for a condition using an AI-enhanced reference like UpToDate, then click through to compare the summary with the linked guideline.
- Registrar / trainee: Use a tool like Glass Health to scaffold a differential diagnosis and draft a management plan, then verify each point against primary sources before presenting to a consultant.
- Pharmacy / therapeutics: Leverage the deep drug-interaction data and comprehensive monographs available within DynaMedex/Dyna AI.
- Cross-cutting: Use an ambient scribe (Doximity's, for example) to capture a consultation, then hand the draft text to a reference tool (Pathway or UpToDate) to find supporting evidence and build a final, well-cited clinical note.
Risks, limits, and mitigations
- Hallucinations & omissions: Always prefer grounded systems with visible citations. A non-negotiable "human-in-the-loop" sign-off process must be in place for all clinical outputs (EBSCO).
- Governance: Any implementation must follow NHS England's CDS guidance, which includes creating a clinical safety case, continuous performance monitoring, and ensuring a problem-oriented design (NHS England).
- Change management: Clinicians require training on prompt discipline and the principle of "verify before you act." Track key metrics like guideline concordance and time-to-answer over a 4–8 week pilot to measure impact.
The strategy question for providers & vendors
For healthcare providers, the goal should be to reduce fragmentation. The best approach is to pilot one primary reference tool alongside one scribe, carefully measuring time saved, citation rates, and concordance with local formularies and guidelines.
For vendors, the message is clear: partnerships beat walled gardens. In the view of iatroX, aggregating trusted sources from multiple providers and exposing transparent, audit-ready citations is the fastest and most effective route to earning clinician trust and achieving scale.