AI for healthcare is already here. It’s just not evenly distributed.

AI is already transforming healthcare, but access remains uneven. Large health systems may have hundreds of AI initiatives underway, while many smaller facilities have yet to deploy their first. For smaller healthcare organizations, concerns over cost, security, and regulation are significant barriers.

AIDEWAY is an open-source platform that brings general-purpose AI capabilities directly into the healthcare facility. It runs locally, allowing organizations to build and deploy AI-based solutions with fixed costs while keeping sensitive data on premises and giving each facility control over how, where, and to what extent AI is integrated into its operations.

We are working with Huntsville Hospital to pilot AIDEWAY for identifying unexpected pathology results and alerting clinicians. Contact Us to discuss early access for your facility.

Request Early Access
Entry 01

Generative AI in hospitals by type.

The best national measure of generative AI in the EHR splits hospitals by teaching status. Major teaching hospitals, the academic medical centers, are twice as likely to be using it today. Nonteaching hospitals are more than twice as likely to have no plans at all.

53.9% Major teaching hospitalsusing generative AI 25.9% Nonteaching hospitalsusing generative AI
Everson et al., JAMA Network Open, 2025; 2,174 nonfederal acute care hospitals, 2024 AHA IT Supplement
Adoption details and key figures

Where each kind of hospital stands on generative AIShare of hospitals by adoption stage, 2024

Using it nowPlanning within a yearLater, never, or unsure
Major teaching hospitalsMajor teaching hospitals: 53.9%53.9%Major teaching hospitals: 25.9%Major teaching hospitals: 20.2%20.2%Minor teaching hospitalsMinor teaching hospitals: 35.0%35.0%Minor teaching hospitals: 30.2%Minor teaching hospitals: 34.8%34.8%Nonteaching hospitalsNonteaching hospitals: 25.9%25.9%Nonteaching hospitals: 20.0%Nonteaching hospitals: 54.1%54.1%0%25%50%75%100%
Everson et al., JAMA Network Open, 2025
14%of the largest health systems have ambient AI documentation fully deployedPoon et al., JAMIA, 2025
64% vs 20%larger health systems vs. $500M to $1B systems piloting or implementing generative AI for revenue cycleHFMA and AKASA, 2025
62.6%of hospitals on Epic had ambient AI documentation by mid-2025, higher with size and marginAJMC, 2026
27% vs 18%health systems vs. outpatient providers with domain-specific AI in placeMenlo Ventures, 2025
46% vs 27%of AI pilots reaching production at large vs. small providersBessemer, Bain, and AWS, 2025
68.8%of independent hospitals have no generative AI plans within a year, vs. 32.4% of system membersJAMA Network Open, 2025
Entry 02

Top reasons hospitals have not adopted generative AI

  • Immature AI tools77%
  • Financial concerns47%
  • Regulatory uncertainty40%
  • Low clinician adoption17%
  • Insufficient expertise or technology14%
  • Lack of leadership support7%
Poon et al., JAMIA, 2025; Scottsdale Institute member systems
Supporting key figures
18%of healthcare organizations say their infrastructure is ready to deploy AI in care deliveryHIMSS Market Insights, 2025
86%of physicians say data privacy assurances are important to their adopting AI; 88% say the same of validated safetyAMA, 2026
3 to 5×the license cost, spent on implementation, validation, and operations over three to five yearsBain
41.2%of rural U.S. hospitals operate at a loss; the national median operating margin for rural hospitals is 2.0%Chartis Center for Rural Health, 2026
63%of health systems call their AI strategy developing or ad hoc; only 44% have an environment to test toolsUPMC and KLAS Research, 2026
2×as likely: rural hospitals depending on their EHR vendor’s models rather than choosing their ownWhitacre, Oklahoma State University, 2026

Open source and local, open models can help.

The performance gap between locally run open models and frontier models is closing. Many everyday tasks, such as extracting information, summarizing documents, and sorting reports, already need less than frontier-level performance. Open source and locally run open models can reduce costs, keep data private, and put your facility in control.

Entry 03

Introducing AIDEWAY

AIDEWAY is an open-source AI hub built for healthcare. It runs entirely inside your facility on hardware you control, keeping sensitive data local and eliminating dependence on external AI services. Built-in access controls, encryption, data governance, and audit logging help support HIPAA compliance and protect patient information. The following are some of AIDEWAY's key features.

  • Data classification

    Identify sensitive information in the data your workflows handle. Configurable classifications help control who can access it, which systems can receive it, and how it must be protected.

  • Workflow engine

    Build custom workflows that connect incoming data, AI analysis, decision rules, and notifications. Reuse steps across solutions and track each execution.

  • Local generative AI

    Built-in integration brings locally hosted models directly into your workflows. Use them to analyze reports and extract information, while controlling which models run and keeping sensitive data on your own hardware.

  • Healthcare connectors

    Connect to electronic health records and other facility systems through FHIR, APIs, file transfers, and email. Bring existing systems into the same governed workflow.

Built by TekFive

AIDEWAY is built by TekFive, a veteran-owned company based in Huntsville, Alabama, and an ARC partner. The team brings nearly 20 years of experience building and supporting systems for agencies including NASA and the VA, with expertise in software development, cybersecurity, and on-premises computing.

Part of ARC's AURA initiative

The AI Research Collaborative (ARC) leads AURA, Accessible Unified Research in AI, connecting open-source builders with domain experts and deployment partners. AIDEWAY is its first project, bringing TekFive and Huntsville Hospital together to develop and validate AI tools around clinical needs.

Huntsville Hospital Pilot

The clinical laboratory testing process is generally divided into three phases: the pre-analytic phase, when tests are ordered and specimens are collected; the analytic phase, when specimens are examined and results are determined; and the post-analytic phase, when results are delivered to clinicians, reviewed, communicated to patients, and acted on when necessary. Unfortunately, communication breakdowns frequently occur in this final phase, and patients may never be notified of important results.

8.2%

Patients said their test results were never communicated

In a survey of more than 213,000 VA patients, 8.2% reported that results from blood tests, imaging, or other diagnostic tests were never communicated to them.

Meyer et al., JAMA Network Open, 2022

29.2%

Abnormal results were not communicated on time

In the same VA study, only 70.8% of abnormal test results were communicated within the required timeframe, leaving nearly 3 in 10 outside the expected communication window.

Meyer et al., JAMA Network Open, 2022

Despite years of mandates and process improvements, gaps in test-result follow-up persist. Alerts, procedural steps, and documentation are intended to reduce risk, but they can also add workload and noise that make important findings harder to recognize.

At Huntsville Hospital, AIDEWAY is being piloted to use local AI to analyze pathology reports and identify unexpected findings that may warrant clinical review, including findings that may not trigger existing rules or safeguards.

Lab Result Received Was ResultUnexpected? Store for Later Processing Notify Clinician Local Model Lab Result Received Was ResultUnexpected? Store for LaterProcessing Notify Clinician Local Model
Example workflow: local AI analyzes a lab report, flags unexpected findings for clinician review, and stores the result for follow-up verification.
AIDEWAY

The path to better care.

AIDEWAY is in its early stages, with an initial pilot at Huntsville Hospital. We'd love to talk if you're interested in early access, have a workflow in mind for your facility, or would like to explore a partnership.

Contact Us