Independent healthcare AI guidance

Choose the right healthcare AI, and roll it out with confidence.

We help hospitals, health systems and practices decide which AI tools to buy and how to put them to work safely. The vendor comparisons, regulation trackers and calculators behind that advice are published free, with our sources on the page. When the decision has to be signed off, we run a fixed fee assessment.

We take no vendor commissions. Our method is published. We sign a BAA before any conversation about patient data.

Diagram: a noisy market of vendors, regulations and evidence read at Clunic's evaluation desk and resolved into a defensible shortlist, named risks and a documented decisionCLUNICEVALUATION DESKTHE MARKETVENDORS24 TRACKEDREGULATIONS10 MONITOREDEVIDENCEPUBLISHEDIN FLIGHTNODESETTLEDWHAT YOU LEAVE WITHSHORTLISTDEFENSIBLERISKSNAMEDDECISIONDOCUMENTED

Where we workEpicOracle HealthathenahealtheClinicalWorksFHIR / HL7HIPAA regulated environmentsNIST AI RMF

What we publish

Three things you can read right now, for free

Vendor comparisons, regulation trackers and calculators. Nothing is behind a form, nothing is paid for by the vendors it covers, and every conclusion shows how we got there.

  • Vendor comparisons

    AI scribes, phone agents and four more categories, side by side on what the integration really costs, what the contract says about your data, and who is left holding the clinical risk.

    6 categories compared

  • Regulation trackers

    HIPAA, the ONC certification rules, the CMS prior authorization rule and the state AI acts. Each page says what the rule requires, who it applies to, and what you have to do about it.

    10 rules monitored

  • Free calculators

    Work out what a scribe would save you, what prior authorization costs you now, and whether you are ready to deploy. Every formula is on the page, so you can argue with the arithmetic before you ever speak to us.

    4 calculators, no signup

The problem

Why most AI pilots never go live

Very few of these projects fail because the software was bad. They fail because nobody owns the decision, clinicians do not trust what the tool produces, and no one will put their name to the compliance answer. All three are fixable, and all three are cheaper to fix before the pilot than after it.

Pilots stall in committee

A vendor demo lands well, a working group forms, and the pilot never leaves the working group. Nobody owns the decision, because nobody agreed in advance what result would count as success. Budget stays open, the clinical sponsor loses interest, and the next fiscal year opens on the same slide.

What good looks like: a success threshold written down before the pilot starts, and one named owner who is allowed to call it.

Clinicians distrust the output

A tool that is right most of the time still costs a clinician time, because they have to check all of it to find the part that is wrong. With no way to see where an answer came from and no way to correct it, redoing the work by hand is the rational choice. Low adoption then gets read as a training problem when it is a design problem.

What good looks like: every output traceable to the source it came from, with an escalation path a clinician can use mid shift.

Nobody will sign off on compliance

The HIPAA Security Rule has no AI chapter, so security review falls back on general principles and every reviewer applies them differently. Vendors answer the questionnaire with marketing material instead of a data flow. The project ends up neither approved nor refused, parked at legal review until the sponsor gives up.

What good looks like: a documented data flow, a signed BAA, and controls mapped to a framework your board already recognises.

Coverage

What we cover so far

Four counts, each of something published on this site that you can open without giving us an email address. The last number is the one that is supposed to stay exactly where it is.

Vendors tracked
24
Regulations monitored
10
Categories compared
6
Vendor commissions
0
Methodology

The CARE Method

How we run the work: chart how the job is done today, architect the tool and the oversight together, run a small pilot, and evaluate it against the numbers you started with before anyone decides to scale. It is a loop, not a straight line, because the evaluation is what tells you whether the design was right.

  1. Chart

    Map how the work is really done today, and work out which parts a tool could pay for.

  2. Architect

    Design the tool, how it connects to your EHR, and the oversight that will hold it, as one plan.

  3. Run

    Start a small pilot with the success target and the stopping rule agreed before it begins.

  4. Evaluate

    Measure against the baseline you took at the start, then decide: scale it, fix it, or stop it.

Solutions

Ten jobs AI tools already do well in a clinic

One page per workflow. Each says what the tool can and cannot do, what it has to connect to, what it saves, and what has to be true before you let it near a patient record.

Why Clunic

Why you can trust what we publish

Clunic started in 2026, so we have no decade of case studies to point at. What we have instead is a method you can check, calculators you can use without talking to us, and a business model that removes the obvious reason to lie to you.

Independent by design

We take no vendor commissions, so the recommendation you get is the one your numbers support rather than the one that pays us best. If the right answer is a product we have nothing to do with, that is the answer you get.

The conditions we work under

Statements about how we operate, not certifications. Each one is something you can hold us to on the first call.

  • BAA-ready
  • FHIR/HL7-native
  • HIPAA-regulated environments
  • NIST AI RMF-aligned

The rules we publish under

No sponsored rankings, no affiliate fees, no paid placement in a comparison, and no number without a source you can open. Our editorial policy states all four in writing, so you can hold any page on this site to it rather than take our word for it.

Read the editorial policy

Everything in writing

Scope, deliverables, findings and fee are documented before work starts, and so is the stopping rule. A recommendation not to proceed is written up with the same care as a recommendation to build, because it is worth more to you.

Services

When you need it signed and dated

The free comparisons and trackers will take you a long way on your own. What they cannot do is score your shortlist against your own systems, sign the finding, or put a date on it. That is what these five engagements are: the same work, done for your organisation, fixed fee, in writing.

  • Vendor Selection and Evaluation

    We score the AI vendors on your shortlist against your clinical, security and integration requirements, then tell you which one to buy. No vendor commissions.

    Fixed fee / Scored vendor shortlist

  • AI Readiness Audit

    A review of your data, systems, staffing and oversight that says which AI tools you could safely put to work today, and which to leave alone.

    Fixed fee / Written readiness report

  • Agent Deployment Roadmap

    A step by step rollout plan that starts with the workflow that pays back first, with the safety checks and the stopping rule agreed up front.

    Fixed fee / Sequenced rollout plan

  • AI Governance and Compliance

    The policies, oversight and audit trails your compliance team needs, written to hold up to HIPAA scrutiny and aligned with the NIST AI Risk Management Framework.

    Fixed fee / Policy and control set

  • Clinical Staff Training and Adoption

    Training for the clinicians and front desk staff who have to use the tool on a real shift, not for a demo audience.

    Fixed fee / Role based curriculum

Insights

All insights

Questions

What buyers ask us first

How do you handle protected health information?

Most of our work does not need PHI. Readiness reviews, vendor evaluations and governance design run on system documentation, observed workflow and de-identified samples. Where a piece of work genuinely requires live data, a business associate agreement is signed first and access is scoped to the minimum the task needs. The forms on this site are not a HIPAA-covered channel, so please keep patient information out of them.

Do you sign a business associate agreement?

Yes. A BAA is available before any conversation that touches PHI, and we will sign yours rather than insist on ours. It costs nothing and it takes the question off the table before the first call, which is why we offer it up front instead of at contracting.

Do we have to replace our EHR?

No. We work inside the system you already run. An AI tool earns its place by taking work out of a workflow you already have, and an EHR replacement would swamp any benefit that tool could produce. Integration is designed against the systems you have now, through the interfaces your vendor already supports, most often FHIR or HL7 v2.

What does an engagement cost?

Engagements are fixed fee, scoped to the size of the organisation rather than billed by the hour. The figure is stated in a written proposal after the evaluation call, once we know what the work actually involves. The call itself is free, and a proposal that recommends less than you asked for is a normal outcome.

How long before we see value?

That depends on which use case goes first and how long your security review takes, so we will not quote a number before we have seen either. What we will do is fix the success threshold and the stopping rule before the pilot starts, so the date you find out is agreed in advance rather than discovered afterwards.

Do you build the tool, or help us buy one?

We help you decide, and we are neutral about the answer. Clunic takes no vendor commissions, so a build recommendation and a buy recommendation pay us exactly the same. For most provider organisations the honest answer is buy it, connect it carefully and watch it closely, and we will say so even where a build programme would have been the larger engagement.