Regulatory / FDA
Ceribell received FDA 510(k) clearance (K261712) for software that adds Alpha Delta Ratio and Relative Alpha trending, burst suppression monitoring and amplitude integrated EEG to its point of care EEG system. The clearance carries no age restriction, so the same trends apply from premature newborns in the NICU to adults in the ICU and emergency department. Ceribell expects to deliver the features through an automatic software update in the first half of 2027.
Our read
One system across neonatal and adult critical care simplifies training and purchasing for hospitals that run both. The features are cleared but not yet shipping, so they belong in 2027 plans rather than this year's.
Product / capability
Version 3.3.7 of the Chartnote Chrome extension, updated 30 September, adds Chartnote Companion. Clinicians can pin a note from their Chartnote list so it stays visible inside any web based EHR, then insert structured sections of that note into the chart with @section shortcuts.
Our read
Clinicians can move scribe output into the chart section by section without switching windows. Because it runs in the browser on any web based EHR, there is no integration project to wait for.
Product / capability
Spring Health's September release embeds the platform in Workday Wellness so employees can reach it in one click from Workday. It also adds homepage cards for work life services and check ins, an add to calendar prompt after booking, and coaching utilization metrics in the Spring Insights employer dashboard.
Our read
Employers on Workday can put the benefit where employees already go. The coaching metrics give HR a clearer view of how that part of the benefit is used.
EHR / interoperability
Caregility and Ascom completed an integration that lets bedside staff call in a Caregility virtual nurse from Ascom Telligence nurse call controls, including the Patient Station, Duty Station and TelliConnect Workflow Station. One button press connects the bedside team to a remote clinician through the Caregility smart room device. Both companies describe it as the first step in a broader agreement, and Ascom joins Caregility's Care Square partner ecosystem.
Our read
Virtual nursing programs stall when staff have to step out of their workflow to ask for help. Putting the request on the nurse call controls already in the room removes that step for Ascom hospitals.
Product / capability
Elsevier launched ClinicalKey Nursing AI, a conversational clinical decision support tool built specifically for nurses. It answers clinical questions at the point of care and shows the evidence behind each answer.
Our read
Most clinical AI assistants were designed around physician questions. This gives nursing leaders an evidence based option shaped around their own workflows.
Model / architecture
Syllable release 26.9.29 adds GPT-6 Sol and GPT-6 Luna to the OpenAI and Azure OpenAI models available to its voice agents. The release also fixes an edge case that caused duplicate outbound campaign calls, improves voice routing during temporary service errors, and adds billing and usage data to Airbyte data feeds.
Our read
The duplicate call fix matters most, since repeated outbound calls reach patients directly. Usage data in Airbyte lets teams track model costs alongside the rest of their reporting.
Product / capability
Enzo Health shipped Clinician Verify for Enzo EHR customers. Every OASIS suggestion from QA or a clinical manager now goes to the clinician who wrote the documentation. On their phone they see the original answer, the suggested one and the reasoning, then choose which answer stays before they sign.
Our read
Home health agencies get a recorded trail of QA suggestions and clinician decisions in place of side conversations. The clinician keeps the final say on their own assessment, which matters when OASIS answers drive payment.
Product / capability
CarePilot's 29 September release adds organization wide controls over note sections, lets clinicians undo a full note pull forward, and shows who wrote a prior note in the Charting Canvas. Billing suggestions for time based codes now follow current Medicare guidance, and the release improves the reliability of the athenahealth integration.
Our read
Practice administrators can now set note structure once for the whole group instead of clinician by clinician. The Medicare update matters most for practices that bill heavily on time.
Clinical evidence
Veracyte reported an analysis of the randomized phase III NRG/RTOG 0815 trial, presented at ASTRO 2026, showing that the Decipher Prostate test predicts which men with intermediate risk prostate cancer benefit from adding short term hormone therapy to radiation. Across 395 patients followed for a median of 10.2 years, those with intermediate or high Decipher scores saw an 8 percentage point absolute reduction in distant metastasis at 10 years with hormone therapy. The 44 percent with low scores saw a 1 point difference that was not statistically significant.
Our read
This moves Decipher from estimating risk to guiding a treatment choice, the kind of evidence payers and guideline panels weigh most. The low score group gained little from hormone therapy, which makes it the group that could be spared its side effects.
Product / capability
Wisedocs launched Legal Intelligence in early access for insurance defense teams. It checks deposition transcripts and demand letters against a claim's verified medical records and history. Each flagged contradiction, gap or admission cites both the testimony and the record, and a portfolio view tracks repeat expert witnesses and provider and law firm relationships across claims.
Our read
Claims and litigation teams usually reconcile the medical file and the legal file by hand, often after reserves are set. Citations on both sides make each flag checkable, which is what a defense attorney needs before relying on it.
Product / capability
Caris launched an enhanced Caris Assure liquid biopsy that combines whole exome sequencing, whole transcriptome sequencing and whole blood analysis in one assay, so tumor derived alterations can be told apart from clonal hematopoiesis and incidental germline findings. Coverage spans more than 23,000 genes at depths up to 40,000x. For patients tested repeatedly during treatment, results are now sorted into molecular response, intermediate or molecular progression, with a median turnaround of about seven days.
Our read
The serial monitoring categories turn a one time profiling test into a way to track whether a treatment is working. Caris says tumor fraction correlates with progression risk, but the release does not publish the validation data behind that claim.
Regulatory / FDA
Tempus received FDA 510(k) clearance for Tempus ECG-MR, software that reads a standard resting 12 lead ECG and flags signs of moderate or severe mitral regurgitation. It is cleared for adults 65 and older with cardiovascular risk factors and no prior mitral valve diagnosis or intervention, and it returns a binary result meant to prompt referral for imaging, not to diagnose. It is the fourth FDA cleared Tempus ECG AI product.
Our read
Mitral regurgitation is often found late because early disease has few symptoms. A screen that runs on ECGs already being taken adds a referral trigger with no new hardware, and it extends the suite that added pulmonary hypertension in August.
Clinical evidence
Artera announced five ASTRO 2026 presentations on its multimodal AI prostate cancer biomarker. One is the first validation in a South Asian population: in the POP-RT trial in India, patients with high MMAI risk had fewer distant metastases with whole pelvis radiation, while intermediate risk patients saw no similar benefit. Another compared risk estimates from commercial testing with STAR-CAP, a database of nearly 20,000 patients, and found them in line with observed outcomes in every NCCN risk group. A third found no evidence of bias by race or age in NRG/RTOG post prostatectomy trials.
Our read
The case for this test has rested mostly on US cooperative group trials. An Indian cohort and a 20,000 patient real world benchmark speak directly to whether the scores hold up outside them.
Safety / governance
Spring Health added a Harm From Others rubric to VERA-MH, its open source benchmark for mental health AI safety, and opened it for 60 days of public comment. The rubric evaluates how an AI system responds when an adult describes risk of physical or sexual violence from another person, scored across five areas using 100 personas, and is published in the VERA-MH repository.
Our read
Open safety criteria give buyers a common yardstick for mental health AI, including tools from other vendors. Buyers can use the draft rubric now to ask any vendor how its AI detects and escalates these situations.
Product / capability
Artera AI agents can now listen for the voicemail beep and start speaking when recording begins, turned on for each agent individually. The same release moves website syncing into an AI agent knowledge base to a background process so it no longer slows live patient conversations, and adds automated tests that check an agent's responses before changes go live.
Our read
Outbound voice agents that talk over a voicemail greeting leave clipped, unusable messages, so beep detection makes reminder and recall calls land when patients do not answer. Pre release response testing is a governance control buyers can ask to see.
Related axis: AI CentralityVendor documentationView source
EHR / interoperability
The 25 Sep Artera Harmony Federal Edition release changes Intake Hub to match patient forms to appointments using the EHR's event ID, so renaming an appointment type in the EHR no longer stops forms from being assigned. Details from a patient's uploaded ID or insurance card now fill in the rest of an Intake Hub form automatically, and the notes say groundwork is in place to write form answers back to the EHR.
Our read
Matching on the EHR event ID removes a silent failure where intake forms stopped going out after an EHR configuration change. Card based auto fill cuts manual entry for patients before the visit. EHR write back is not yet live.
Product / capability
Insilico announced that Model Context Protocol servers are now available across its Pharma.AI software, so outside AI agents can connect directly to its biology, chemistry and biologics engines, including Generative Biologics, PandaOmics and Chemistry42, and run multistep discovery workflows. The announcement previews a 30 September webinar that will present new PandaOmics Agent skills and other platform updates.
Our read
MCP access lets a customer's own agents drive Insilico's engines, which turns Pharma.AI from a set of applications into a component of the customer's stack. Teams should ask how data sent through those servers is logged and whether it is used to train Insilico's models.
Product / capability
Corti's Guided Generation now passes user edited, user added and discarded facts to the model with explicit type metadata. Edited facts take precedence over the transcript where the two conflict, and discarded facts are excluded from the note even when the transcript mentions them. Previously, edits and discards were not reliably reflected in the generated note.
Our read
This closes a common gap in ambient documentation, where a clinician's correction is undone by the next draft. Developers building on Corti should retest note generation, since conflicts now resolve in the clinician's favor rather than the audio's.
Product / capability
Veeva announced Study Builder Agent, which configures Veeva EDC forms, visits and edit checks and Veeva DQS listings directly from a study protocol, reusing an organization's existing standards and the CDISC USDM model, and generates test data. Veeva says it can bring study configuration down to as little as one day. It is planned for early adopter availability in December 2026, is included in EDC with no additional license, and is delivered as a Claude Cowork plugin.
Our read
Study build is one of the slowest steps in trial startup, so the one day claim is worth testing on a real protocol once early adopter access opens. Sponsors should ask which study designs the agent handles and what a person still reviews before a build goes live.
Related axis: AI CentralityVendor announcementView source
Model / architecture
InsightRX released Gemini 2026.2, an update to the model selection engine in InsightRX Nova that adds a new general adult vancomycin model, McCarthy 2026, plus CKD-EPI eGFR versions of the published models that improved with that input. McCarthy 2026 was fit on a stratified sample drawn from 549,171 patients across 304 US organizations, scales body size on fat free mass and bends the eGFR effect on clearance at 90 mL/min/1.73 m2. InsightRX reports that the release raised individual prediction accuracy from 66% to 68%, cut bias (MPE) from 1.8% to 0.9% and cut error (NRMSE) from 28% to 27% versus Gemini 2026.1.
Our read
Health systems using Gemini get the new model recommendations automatically, and InsightRX states the Hughes 2024 obesity model is now mostly selected only for patients with obesity. It also shows the vendor retraining and revalidating models on network data after deployment. The accuracy gains are vendor reported and modest.
Product / capability
Helix launched Helix Research Workspaces, a Trusted Research Environment inside Helix Explorer built on the Databricks Data + AI platform, giving approved partners governed compute, storage and analysis tools over more than 550,000 GenoSphere clinicogenomic records (Exome+ data linked to 13+ years of EHR history plus claims). It ships a genomic analysis toolkit for GWAS style, PheWAS and gene burden studies, an AI coding assistant Helix describes as trained on GenoSphere, no code data browsing, project notebooks with built in data access controls and monitoring, and real time job status and compute cost tracking.
Our read
Pharma and health system research teams can run population scale genomic analysis inside a Helix provisioned, access controlled workspace instead of standing up their own infrastructure, which changes the deployment and data governance model for GenoSphere licensees beyond the earlier Cohort Builder.
Product / capability
Kosmos can now read licensed full text research from specialist journals across Springer Nature's Nature portfolio when it generates hypotheses or evaluates targets, under a subscription agreement Edison announced on 24 September. Kosmos links each surfaced article to its Version of Record, so outputs reflect corrected versions including retractions and editors' notes. Edison states the Nature portfolio is accessible in Kosmos as of the announcement date and calls it the first of several content partnerships.
Our read
R&D teams evaluating an AI scientist need to know which licensed literature grounds its conclusions, and retraction aware Version of Record linking reduces the risk of hypotheses built on withdrawn findings.
Regulatory / FDA
The FDA Molecular and Clinical Genetics Devices Panel voted on GRAIL's Premarket Approval application for the Galleri multi cancer early detection blood test, for screening adults aged 50 and older. The 10 voting members were unanimous on safety, voted 6 to 4 on effectiveness, and voted 7 to 2 with one abstention that benefits outweigh risks. FDA is not bound by the vote and GRAIL expects a final PMA decision in the coming months.
Our read
Galleri is the first multi cancer early detection test to reach a PMA panel, and approval would change its regulatory standing for payers and health systems weighing coverage and adoption. The split 6 to 4 effectiveness vote is the risk to watch before FDA decides.
Regulatory / FDA
Aidoc received FDA Breakthrough Device Designation for CARE Chest X-Ray Triage, built on its CARE foundation model, covering triage of pneumomediastinum and lobar or lung collapse. It is Aidoc's third breakthrough designation in just over a year, after CARE Multi-Triage Body CT and CARE First Read for Chest X-Ray. The 510(k) is under FDA review and the product is not yet available for sale in the United States.
Our read
A breakthrough designation can speed FDA review but does not clear the product. Radiology groups interested in foundation model triage for chest films should treat this as a roadmap signal and ask Aidoc for expected clearance timing.
EHR / interoperability
voize's voice documentation for nurses now writes directly into ORBIS U, the hospital information system from Dedalus HealthCare. Nurses speak observations at the bedside, voize structures them, and the entries land in the ORBIS record where the rest of the clinical team can use them. voize says Agaplesion already runs the integration in production and that after a three month pilot it saves up to one hour of documentation per nurse per shift.
Our read
For German hospitals on ORBIS, voice documentation for nursing no longer means a second system to reconcile. The one hour figure comes from a single customer pilot, so treat it as a target to test rather than a benchmark.
Product / capability
Syllable release 26.9.22 added Insights Reports, a workflow detail view for reviewing analyzed results, configuration and session context, and Experiments, which splits live traffic across weighted voice agent variants and compares the results by session. The release also fixed voicemail delivery for carriers with short recording windows.
Our read
Experiments turn agent changes into measured tests on real calls. Health systems should agree on the outcome being tested, such as completed scheduling, and on how patients are informed, before splitting live patient traffic between variants.
Product / capability
Relatient extended the self scheduling and call center scheduling tools in its Dash platform to behavioral health providers, adding secure self service booking and automated appointment management. Relatient cites one behavioral health customer where 60 percent of self scheduled appointments came from new patients and 47 percent of booked appointments were seen within three days.
Our read
Behavioral health scheduling carries privacy and intake requirements that general scheduling tools handle poorly. Organizations should ask which intake steps are automated and how sensitive appointment types are kept out of reminders and call center scripts.
Model / architecture
Deciphex launched CipherX, a pathology AI engine that pairs foundation models with a semantic layer translating their representations into named, pathologist validated tissue signatures. Deciphex states CipherX is in production in its Diagnexia clinical diagnostic service and its Patholytix research business, and reports negative predictive values of 99.85 percent for adenocarcinoma and 98.76 percent for melanoma in production, with every Diagnexia case signed out by a pathologist.
Our read
Labs using Diagnexia get CipherX inside the service rather than as a separate purchase. The predictive values are vendor reported, so ask for the case mix and prevalence behind them and which signatures are used in clinical sign out today rather than research only.
Clinical evidence
Cohere Health published performance results for Cohere Capture, which extracts HEDIS gap closures and denominator exclusions from clinical data a health plan already holds, including prior authorization notes, imaging reports and EHR data. Cohere reports an average 15 percent improvement in Stars performance per measure, 107 gaps closed per 1,000 members with records, and a minimum 5 to 1 return on a pay for performance model.
Our read
These figures are vendor reported and the release does not describe how they were measured. Plans should ask for the method and baseline behind the 15 percent figure, and how much of the lift comes from prior authorization records Cohere already sees as the plan's utilization management vendor.
EHR / interoperability
A Voicebrook VoiceOver PRO integration with the HALO AP platform is now available, adding voice control of case and slide navigation (zoom, pan, annotation, tab switching) and voice ordering of stains and AI analyses. A single command pulls HALO Clinical AI biomarker results such as ER and PR percentages into the matching CAP cancer protocol checklist, and sites that report inside HALO AP can pass results through to the LIS.
Our read
Labs running HALO Clinical AI no longer need a pathologist to re key biomarker values into the synoptic report, which removes a transcription error point in a clinically weighted field. HALO Clinical AI remains Research Use Only in the US, so automated result capture applies to clinical reporting only where the product is cleared or marked.
Model / architecture
Ambience described a change to its clinical speech recognition pipeline from a mini batch ensemble to a rolling window ensemble, which keeps recent audio in scope so a reconciliation model can revise earlier transcript segments as more audio arrives. Ambience reports keyword error fell 16 percent against the previous design on a 226 encounter evaluation spanning 28 specialties and 15 clinicians, at the cost of a median processing cycle roughly 40 percent longer.
Our read
Errors on drug names, dosages and negations are what make an ambient note unsafe to sign, which is why keyword error is the figure to watch. The evaluation is the vendor's own and spreads 226 encounters across 28 specialties, so ask whether the gain holds in your highest volume specialties.
Product / capability
Enzo shipped Handoff Summary, a pre visit screen that brings together the patient's context and clinical focus, the alerts and requirements for that visit, and, for Enzo EHR customers, every care plan change since the last visit. It opens in one tap from the schedule or the patient profile and keeps earlier summaries attached. It is available to both Enzo Scribe point solution customers and Enzo EHR customers.
Our read
Home health agencies evaluating a scribe can see the product now covers visit preparation as well as documentation. The care plan change view only applies to agencies running Enzo EHR, so Scribe only buyers get a narrower version.
Related axis: AI CentralityVendor documentationView source
Clinical evidence
Doximity published Bedside Bench, an open source clinical AI benchmark of 500 cases across 10 sub benchmarks, including drug safety, guideline adherence, landmark trials, hallucination and false premises, health equity and diagnostic safety. The 250 case training split and the grading rubrics are public on Hugging Face, and the test split is held out. Bedside Bench is one of the domain benchmarks in Fireworks' Specialized Intelligence Index, and Doximity states that Doximity Ask ranked first against frontier models when Fireworks graded it.
Our read
Buyers get a public rubric they can run against Doximity Ask and competing clinical reference tools. The ranking is stated without published scores and the benchmark was written by Doximity, so ask for the per benchmark numbers before relying on it.
Clinical evidence
Nature Medicine published PATHFINDER 2, the registrational US study of Galleri, with 35,878 enrolled and 32,007 participants analyzable for performance. The test showed a 0.54% cancer detection rate, 60.3% positive predictive value, 99.64% specificity, 39.3% episode sensitivity for all cancers (69.8% for a 12 cancer subgroup) and 91.3% cancer signal origin accuracy. Of 35,335 safety analyzable participants, 0.6% had invasive procedures and no serious study related adverse events were reported.
Our read
These are the peer reviewed performance and safety figures behind the PMA, giving buyers a published US intended use benchmark for follow up workload and false positive burden. Results were first presented at ASCO in May 2026, so the change is the peer reviewed publication.
Product / capability
Atropos Health states it expanded its federated Atropos Evidence Network to cover over 330 million U.S. patients, including more than 150 million EHR records with clinical notes, and added imaging and omics data linked to clinical outcomes and mortality through data partners including Norstella, Arcadia and OneMedNet. The network is queried through ChatRWD, Green Button and GENEVA OS, with the Real World Fitness Score grading data quality per question. The company claims temporal queries run up to 50 times faster and 30 times cheaper than traditional analytics; no method for that comparison is published.
Our read
Health systems and life sciences teams generating real world evidence can now study questions that need imaging or genomic context alongside EHR and claims data. The speed and cost figures are vendor claims without a stated baseline, so ask for the comparison behind them.
EHR / interoperability
Cynerio's medical device security now ships inside Axonius for Healthcare, and cynerio.com redirects there. Axonius release 9.0.7 lets the TRIMEDX adapter connect through the TRIMEDX Client Inventory (v5) API alongside the older MDSP Inventory API, and adds a Network Inspectors page to the Cyber-Physical Assets module for provisioning and tracking virtual network inspector appliances by site.
Our read
Health systems whose clinical engineering inventory runs through TRIMEDX can connect on its current API and keep biomedical asset and security records in one inventory. The inspector page is an administrative convenience for sites using passive network sensors.
Clinical evidence
Nature Medicine published Galleri test performance from the randomized NHS-Galleri trial across three annual screening rounds in England. Specificity ranged from 99.50% to 99.60% and cancer signal origin accuracy from 91.1% to 93.6%, while positive predictive value fell from 58.0% to 50.4% to 45.8% across rounds. Episode sensitivity was 26.7% to 37.2% for all cancers and 47.6% to 63.4% for 12 prespecified types.
Our read
This is peer reviewed randomized trial evidence in an intended use population, and the falling predictive value in later rounds bears on the cost of repeat annual screening. The data were first presented at ASCO in May 2026, so the change is the peer reviewed publication.
Product sunset
ClosedLoop, the healthcare data science and predictive analytics company, no longer sells a healthcare product. Its domain now offers a workspace for team based agentic software development, with no healthcare product, page or customer route. No acquirer of the healthcare business has been announced. The date the healthcare offering ended is not published; the change was observed on 21 September 2026.
Our read
Organizations still running ClosedLoop models for readmission, utilization or risk stratification should confirm in writing who supports those models now and what happens to their data, because the domain no longer publishes a healthcare support route. ClosedLoop no longer appears in the index's value based care lane; buyers comparing risk stratification vendors will find current options on that lane page.
Vendor documentationView source
Model / architecture
Iambic released Enchant v3, the third generation of the multimodal transformer behind its drug discovery platform. Iambic states the model has 41 billion parameters, was pretrained on 4.5 trillion tokens, and covers more than 6,000 molecular properties across 16 biomedical data modalities using a mixture of experts architecture. Iambic says it will be used on internal and partner programs.
Our read
Enchant v3 is the model partners get when they work with Iambic, so its prediction accuracy is the thing being bought. Iambic says scaling has held across three generations, including on endpoints with sparse data. Ask for the benchmark results in the Enchant v3 report and how they hold on targets with little training data.
Security / compliance
Miya Precision 7.11.0 introduces a three tier restricted access model for patient records: full access, break glass access that requires justification via an individual PIN with audit tracking, and no access. Alcidion states the design supports NHS safeguarding principles.
Our read
Privacy and information governance teams get record level access control with an audited break glass path, which matters for restricted and safeguarding sensitive patients.
EHR / interoperability
Alcidion's Miya Precision 7.11.0 adds a native inpatient admissions workflow, completing its admission, transfer, leave and discharge set, with one way or two way integration to the Patient Administration System so clinicians no longer need to open the PAS. The release also adds Miya Vitals, for capturing observations on charts defined per care area such as ICU, ED and general wards with validation checks and warning thresholds, and Dynamic Fluid Balance Management, which calculates fluid restriction limits from patient weight.
Our read
Trusts running a separate PAS can manage the full inpatient encounter in Miya Precision while keeping the PAS in sync, which cuts double entry. Vitals capture is manual entry only in this release; Alcidion states capture from bedside monitors is planned for a future release.
Product / capability
Upheal added voice input to Upheal Assistant, so clinicians can speak requests instead of typing them. Reports, still in beta, can now be downloaded as an image or a spreadsheet containing headline figures and full chart values, and any report can be opened in the Assistant for questions about it.
Our read
The Assistant now reaches practice reporting as well as session notes. Practices that hand report data to bookkeepers or billers should check what the spreadsheet export contains, since that data now leaves the platform in one step.
Related axis: AI CentralityVendor documentationView source
Security / compliance
Zus refreshed the code sets behind its automatic tagging of specially regulated records against current VSAC value sets. Contraceptive management records now receive the SEX specially regulated label alongside abortion and gender affirming care, the categories covered under California AB 352. Data customers had already written to Zus was retagged under the updated sets.
Our read
First party records that match these codes also receive the Restricted label, which Zus says withholds them from other builders, from Lens summarization, and from CommonWell and Carequality responses. Third party records get the specially regulated label only, so customers should confirm their own downstream systems honor it.
Regulatory / FDA
DeepHealth received FDA 510(k) clearance for Chest XRay, a computer aided detection tool built on a foundation model that detects and localizes abnormalities on chest radiographs. It is the next version of the technology DeepHealth acquired with Gleamer, it is commercially available in the US now, and existing customers are eligible for the update.
Our read
A foundation model clearance matters because one model covers many findings, rather than one narrow algorithm per finding. Radiology groups should ask exactly which findings the 510(k) covers, since marketing for foundation models tends to run ahead of the cleared indications.
Product / capability
Upheal added a Reports section with a curated report library and five thematic dashboards for tracking practice metrics. Users can also ask the Upheal Assistant to generate a specific chart.
Our read
Small therapy practices usually export to spreadsheets to see how the business is doing. Charts generated on request are convenient; check that they draw on the same numbers as the fixed dashboards before using them for payer or staffing decisions. Practice reporting is a convenience layer, not a clinical capability, so it should not move an evaluation far on its own.
Related axis: AI CentralityVendor announcementView source
EHR / interoperability
Assort Health expanded its direct workflow automation for practices on NextGen Enterprise. Using NextGen's Platinum API tier, Assort's AI agents can read and write appointment slots, referrals, clinical notes, chart alerts, diagnoses, procedures and payment workflows directly in the EHR.
Our read
Write access is the line between an agent that answers the phone and one that finishes the job. Practices should confirm which of those write actions are switched on by default, and how each one appears in the NextGen audit trail.
Product / capability
RadiantGraph launched Care Guidance Agents, AI voice agents that give patients with chronic conditions ongoing support: education after a diagnosis, daily management coaching and preventive care, with dedicated agents for musculoskeletal and behavioral health. RadiantGraph also says it has been approved for the CMS ACCESS Model, so participating providers can enroll Medicare Part B patients.
Our read
ACCESS pays for outcomes, not activity, so any vendor in it has to show that its agents actually move blood pressure, pain or depression scores. Ask RadiantGraph which ACCESS tracks it covers and what outcome data it can hand back to a provider.
Product / capability
Sorcero's Fall 2026 release adds purpose built AI solutions for field medical, medical communications, and therapeutic leadership and launch teams, plus a Medical Strategy solution for tracking scientific resonance. It also introduces Sorcero Wizard, a conversational layer for working with the data, and analytics views that show data ingestion and intelligence generation in real time.
Our read
Medical affairs teams work under strict rules about what they can say and to whom, so governance is the feature that matters here. Ask how the platform separates what field medical can see from what commercial teams can see.
Related axis: AI CentralityVendor announcementView source
Security / compliance
Upheal now lets practices invite non clinical staff such as billers and practice managers. Those roles cannot see clinical notes, treatment plans, transcripts or AI generated clinical content.
Our read
Therapy notes carry some of the strictest privacy expectations in healthcare, and giving a biller full access just to reach the invoices was a real exposure. Role based access that walls off the AI generated clinical content is the right default.
Clinical evidence
Knowtex reported results from a US Department of Veterans Affairs assessment of its ambient scribe at the Kansas City VA, covering 18 primary care providers: most saved one to two hours of after hours work, and patient experience scores rose nearly three points to 95.8 percent. In the same announcement Knowtex launched a Frontier AI Lab for Healthcare and KnowBench, an internal benchmark on which it scores its own platform at 97.99 percent.
Our read
The VA assessment is the part worth weight: an outside health system, working clinicians and patient scores. KnowBench is Knowtex testing its own product, so treat that figure as marketing until someone else runs it. Eighteen providers at one site, reported by the vendor rather than published, is too small to carry an evaluation on its own.
Product / capability
AliveCor launched KardiaACCESS, its program under the CMS ACCESS Model, the 10 year Medicare initiative for technology supported chronic care. Eligible Original Medicare beneficiaries get clinician guided care and AliveCor's connected devices for managing blood pressure, diabetes and heart health, at no out of pocket cost.
Our read
This is AliveCor becoming a care provider under Medicare, not only a device maker. Clinicians referring patients should understand who is accountable for the outcomes ACCESS pays on, and how AliveCor's clinicians coordinate with the patient's own doctor.
Acquisition / corporate
Hello Patient acquired Converse Health, adding AI agents that handle the document and chart work around a visit. The combined platform reads incoming faxes, matches or creates the patient record, files documents in the EHR and then follows up with the patient by call or text. Hello Patient integrates with ModMed, athenahealth and eClinicalWorks.
Our read
Most practice automation stops at the phone, while the paperwork behind a referral still lands on staff. Buyers should ask how the agent handles a fax it cannot match to a patient, because a misfiled clinical document is a patient safety problem, not just an admin one.
Regulatory / FDA
AZmed received its sixth FDA clearance, for AZchest to detect and localize pneumothorax and pleural effusion on chest radiographs. That brings its Rayvolve suite to eight FDA cleared findings, and AZchest can now flag up to five findings on a single chest image.
Our read
Pneumothorax is a time critical miss, which is why this clearance matters more than the count. AZmed cites reading time reductions of up to 31 percent for readers outside thoracic radiology; that is the group to test it with, since specialist readers gain least.
Acquisition / corporate
Andros was acquired by Medallion. Andros customers will move onto Medallion's AI credentialing platform, which adds automated primary source collection, AI agents for provider outreach and real time status tracking.
Our read
Health plans that chose Andros for its NCQA certified verification now have a new platform underneath it. Ask for the transition timeline in writing, and whether NCQA certification carries across without a gap.
Product / capability
Abridge launched pre bill review for clinical documentation integrity, coding and revenue cycle teams. It checks inpatient claims before submission, comparing the coded diagnoses and DRG against what the clinical record supports, including what was captured in the encounter, and surfaces discrepancies with the evidence. Abridge does not change documentation, codes or claim status; the CDI team decides what to hold, correct or release.
Our read
This takes Abridge from the exam room into the revenue cycle, where CDI vendors have worked for years. The advantage it claims is the encounter itself: a coder can see what was actually discussed, not just what was written. Ask how discrepancies are ranked, and how often a flag leads to a correction rather than a physician query.
Product / capability
Oracle made its Clinical AI Agent available to inpatient nurses in the US. Nurses get voice driven chart navigation, acute nursing summaries and voice enabled discrete charting inside the Oracle Health Foundation EHR.
Our read
Ambient AI has so far been sold to physicians, while nurses do more of the documentation and have had less help. Discrete charting by voice is the piece to test, because structured data that lands in the wrong field is harder to catch than a messy note.
Acquisition / corporate
Veracyte acquired Convergent Genomics, adding the UroAmp urine tumor DNA test to its urology portfolio. Veracyte can now offer genomic testing from urine, tissue and blood across bladder cancer care.
Our read
A urine test can help monitor bladder cancer between cystoscopies, the procedure patients most want to avoid. Urologists should ask how UroAmp results will be reported alongside Veracyte's existing tests, and whether ordering stays the same during the integration.
Product / capability
Aignostics released PathoSearch in early access, a visual search engine for pathology: from a screenshot of a region of interest on an H and E slide it retrieves morphologically similar, diagnosed reference cases so a pathologist can build a differential in minutes. Search runs on embeddings from Atlas 2, Aignostics' pathology foundation model co developed with Mayo Clinic, against a curated multi center reference set of more than 310,000 whole slide images from over 35,000 cases spanning more than 300 diagnostic entities across 27 organs, covering thoracic, digestive, soft tissue and bone, female genital, and urinary and male genital cancers at launch. The first integration is live inside Techcyte's Fusion AP slide viewer, where a selected region is passed to PathoSearch without export or upload. PathoSearch is free during early access, by waitlist, and is labeled research use only, not for diagnostic procedures.
Our read
Rare and complex cases are where a pathologist reaches for a textbook or a paid consult, and image similarity search against a diagnosed reference set is a plausible replacement for the first of those. The research use only label is the operative constraint: this cannot be part of a signed out diagnosis today, and a lab evaluating it should treat the Fusion AP integration as the way to try it inside the workflow rather than as a clinical deployment.
Related axis: AI CentralityVendor announcementView source
Product / capability
Parlance made its IT help desk intelligent virtual assistant available on the ServiceNow Store, alongside a new identity verification step for the voice channel: for sensitive requests such as a password reset, the assistant reads a one time code to the caller, who confirms it in Cisco Duo or another enterprise authenticator before the action proceeds. Parlance cites industry estimates that password resets drive 20 to 50 percent of health system IT help desk call volume.
Our read
The Duo step is the part with security consequence: an unverified push or an SMS code confirms that someone approved something, while a code read on the live call and confirmed in the authenticator ties the approval to that caller and that request, which closes the social engineering path that help desk voice bots opened. Health systems on ServiceNow can now evaluate this from the Store rather than through a custom integration.
Model / architecture
Maverick Medical AI announced a proprietary Clinical to Revenue Foundation Model that now powers its two products: CodeAgent, which engages physicians at the point of care to flag missing documentation, coding opportunities, medical necessity gaps, modifier errors, MIPS requirements and payer specific risks before a note is finalized, and mCoder, which generates billing ready codes from that clinical context while further agents validate payer policy and compliance rules before a claim goes out. Maverick says the model learns revenue cycle intelligence common across specialties while adapting to specialty specific workflows, and reports 85 percent or better direct to bill performance in production.
Our read
A vendor claiming a foundation model for coding is making a model disclosure claim, and the release gives a buyer nothing to check it against: no architecture, training data, evaluation set or specialty breakdown. The direct to bill figure is the number to test on the institution's own claims mix, and the CodeAgent behavior is the one to watch clinically, since a system that prompts physicians for coding opportunities at the point of care sits close to the line compliance teams draw around upcoding.
Clinical evidence
Primaa and PathPresenter announced completion of their joint validation study of Cleo Skin, Primaa's dermatopathology AI, running inside PathPresenter's FDA cleared clinical viewer. The study covered a realistic case mix of melanoma, naevus, squamous cell carcinoma, basal cell carcinoma and benign lesions, and the companies report faster and more consistent diagnostic reporting when the AI populates structured reporting fields inside the viewer. Interim phase one results PathPresenter published in March reported a 25 percent reduction in analysis time and a 6 percent improvement in diagnostic accuracy with the model, with higher interobserver agreement on lesion classification.
Our read
The finding worth holding onto is that the gains came from the AI being inside the viewer rather than beside it, which is the deployment shape most labs still have not achieved. The evidence is vendor and partner reported, with the population and endpoints described at summary level, so treat it as a workflow study rather than a diagnostic accuracy trial until the full results are published, and ask which of the interim numbers held at completion.
EHR / interoperability
Canary Ambient for Physicians is generally available inside Microsoft Dragon Copilot, listed on the Microsoft Marketplace as part of the Dragon Copilot AI Apps and Agents program Microsoft opened on 21 August. Inside the Dragon workflow, Canary analyses the acoustic and linguistic properties of the recorded encounter to surface screening signals for mild cognitive impairment and Alzheimer's disease (Canary Cognitive) and for depression and anxiety (Canary Behavioral), with no separate application or added step during the visit. Canary describes the output as clinical decision support for a clinician to interpret, not a diagnosis, and states that Marketplace listing is not self service: enablement still goes through Canary for licensing and deployment.
Our read
The change is distribution, not the model: a health system already on Dragon Copilot can now procure and deploy voice biomarker screening through a channel its IT and legal teams have already approved, which is the step that has kept this category in pilots. The questions Canary itself says to ask are the right ones, and they belong in the contract: intended use, patient consent for biomarker analysis, audio retention and deletion, whether encounter audio trains vendor models, and performance across accent, age and language.
Product / capability
IQVIA launched Predictive Clinical Development, a suite covering site selection and study design, an automated start up path it calls Push Button Start-Up, and real time data cleaning aimed at shortening the gap to database lock.
Our read
Each of the three targets a different well known delay, and the one worth interrogating is real time data cleaning, because database lock timelines are usually held up by query resolution rather than by query generation. Sponsors evaluating this should ask which of the three has a measured effect on an actual study rather than a modeled one.
Related axis: AI CentralityVendor announcementView source
Product / capability
BrainCheck added BrainCheck Screen, a pre screening tool that identifies which patients need fuller cognitive testing. The vendor reports validation on a 99 patient dataset showing 92 percent sensitivity and 74 percent specificity for detecting cognitive impairment.
Our read
A stated population and both operating characteristics is more than most vendors in this index publish, and it is enough to reason about. Sensitivity of 92 percent with specificity of 74 percent is the right shape for a pre screen, since the job is to avoid missing people and to accept that a quarter of those flagged will test normal. The dataset is small at 99 patients, so the number to ask about is performance in a primary care population rather than a validation cohort.
Product / capability
Clearwave connected its Scheduling Agent to Reserve with Google, which puts a Book Online button on a practice's Google Business Profile. Patients can book from Google Search and Maps without being handed off to a separate site.
Our read
Patient acquisition for most practices happens on a search results page, and every redirect between that page and a confirmed appointment loses people. Removing the handoff is a small change with a measurable outcome, and it is measurable, which makes it worth asking the vendor for booking completion rates before and after.
Product / capability
Lumea integrated TreatmentGPS's Alpine Path into its digital pathology platform, embedding treatment guidance into the pathology report itself. For prostate and bladder biopsies the report now carries NCCN based risk stratification, ancillary testing recommendations and a plain language summary written for the patient.
Our read
This moves the pathology report from a finding to a recommendation, which is a genuine widening of what a pathologist is publishing and worth a governance conversation before it is switched on. The plain language summary is the quietly significant part, since it means the report is now written for two audiences with different tolerances for uncertainty.
Regulatory / FDA
CMS approved a New Technology Add-on Payment for Bayesian Health's FDA cleared continuous AI sepsis monitor, creating Medicare payment support for health systems running the platform.
Our read
This is the rarest event in this index, which is an AI tool acquiring a reimbursement pathway rather than only a clearance. Clearance establishes that a product may be used and says nothing about who pays for it, and the absence of a payment mechanism is the reason most cleared clinical AI stalls after the pilot. An NTAP moves the conversation from the innovation budget to the operating one.
Acquisition / corporate
LeanTaaS acquired Aidin and is folding discharge readiness intelligence and post acute placement into iQueue for Inpatient Flow, extending the platform from admission through transition out.
Our read
LeanTaaS has sold capacity optimization inside the hospital walls for a decade, and the constraint it kept hitting was a bed that cannot be freed because a post acute placement has not been arranged. Buying the placement workflow rather than integrating with one is a statement about where it thinks the remaining length of stay savings are.
Model / architecture
Insilico released a set of small language models trained as scientific specialists for chemistry and biology through its MMAI Gym for Science framework. The set covers chemical synthesis, ADMET prediction and potency prediction across GPCR and kinase panels, and includes a single step retrosynthesis model built on Liquid AI's 2.6 billion parameter architecture.
Our read
Naming the base architecture and the parameter count is unusual in this category and is the reason this is graded on model transparency rather than capability alone. For discovery teams the practical read is that these are narrow specialists rather than a general assistant, which is the right shape for ADMET and potency work where a confident wrong answer is expensive.
Product / capability
bioAffinity extended CyPath Lung from early detection into surveillance of lung cancer survivors. The noninvasive sputum test is now positioned to complement CT imaging when monitoring for recurrence and when evaluating new or changing pulmonary nodules in patients who have completed curative intent treatment.
Our read
Surveillance is a larger and more repetitive population than initial screening, so this is a meaningful widening of where the test can sit in a pathway. The specific job is risk stratification when imaging is equivocal, which is the moment a pulmonologist currently has to choose between watchful waiting and an invasive procedure on incomplete information.
Acquisition / corporate
LeanTaaS acquired Aidin and will connect its care transition workflows to iQueue for Inpatient Flow. Aidin's post acute placement workflow now sits downstream of LeanTaaS's discharge readiness intelligence rather than alongside it.
Our read
Anyone currently evaluating Aidin is now evaluating LeanTaaS, with the contract, roadmap and support relationship that implies. The combined product is a bet that discharge delays are a capacity problem and a placement problem at the same time, which is true in most hospitals and is why the two have historically been bought separately and blamed on each other.
Product / capability
Xsolis launched GenAI Peer-to-Peer Clinical Synopsis as an add on inside Dragonfly Advise. It reads longitudinal EMR data including labs, vitals and nursing notes, and generates a structured synopsis highlighting severity indicators and medical necessity rationale ahead of a peer to peer denial review.
Our read
The vendor puts the manual version of this task at 15 to 20 minutes of chart reconstruction inside a 24 to 72 hour regulatory window, which is a specific enough claim to test against a physician advisor's own calendar. The risk to check in evaluation is what the synopsis omits, since a medical necessity argument built from a summary is only as good as the exclusions nobody sees.
Related axis: AI CentralityTrade pressView source
