Evaluating ClinicalSim
Last updated: October 2026
Use this guide to share the product's intended use, rollout requirements, and limits with the people reviewing it. Start with the summary, then go straight to the topic your team needs.
For participants
Spoken practice with AI patients, followed by feedback that shows strengths and what to practice next, with the transcript behind each score.
For leaders
Cohort progress, recordings, transcripts, and feedback under the program's access rules, so leaders can plan coaching and further practice.
To launch
Start with a communication priority, a participating group, and a named standard. Participants use a browser, and IT may need to allow the voice service.
Before approval
Agree on access, retention, and reporting rules. Scores support formative practice and program judgment, and must not drive employment or other high-stakes decisions.
Five complete AI patient simulations are published with audio, transcript, and the full report, with no sign-in required. Each shows strengths, areas for focused practice, and the participant's words behind the feedback. See the examples.
1. What it is for, and what it is not for
ClinicalSim is intended for spoken communication practice and formative assessment. Each simulated conversation produces feedback tied to a named standard and the transcript, so participants can see what worked and what to practice next. It produces no patient-facing output, makes no diagnostic or treatment recommendation, and creates no clinical documentation.
It extends a standardized patient program rather than replacing one. SP encounters stay the high-stakes assessment and the place live coaching happens. ClinicalSim adds the repetitions between them.
It does not price malpractice risk, and it does not benchmark one institution against another. For the idea behind the product in one page, read how participants practice and leaders review results.
It is audio only. Voice practice surfaces pacing, silence, word choice, and how a clinician responds to emotion in the moment. It cannot assess eye contact, body language, or physical presence, so those stay with live encounters and human review.
2. Who takes part in an enterprise review
Communication training often involves different teams. Program or health system leaders define the need, finance reviews the budget, IT and privacy teams review data handling, and clinicians or learners use the product. These pages explain what each group needs to know.
- Program directorsGive each learner a clear next step between coaching sessions
- DIOs and GME leadershipOne reporting structure across programs, with local clinical standards
- Simulation center directorsExtend your SP program, don't replace it
- Clinical competency committeesRubric-scored evidence for CCC review
- Medical school and UME leadershipSequence communication across all four years
- Faculty and clinician educatorsPractice the conversations faculty are expected to model
- Risk and patient safety leadersCommunication factors in 40% of asserted malpractice cases
- Quality and patient experience leadersPractice the service standards your institution already teaches
- Nursing leadershipNurses practice the conversations behind patient experience, retention and safety
3. What the evidence establishes, and what it does not
The communication frameworks ClinicalSim applies were built for trained human raters observing real encounters, and that is the setting in which their published reliability was established. Scoring them with AI in a simulated encounter goes beyond that setting, so a framework's reliability does not carry over to a ClinicalSim score. Each score is a formative signal backed by verbatim transcript evidence, and it should not carry a high-stakes decision on its own.
We don't claim our scoring is more accurate or more valid than a faculty member's read. We don't have the validation data to say that, and we won't claim it until we do. How cases are built and scored is documented on our methodology page.
Studies of ClinicalSim are underway at partner institutions, and we will not publish an outcome from any of them until that study's owners have confirmed the record in writing. What is public today is the conference presentation list on our research page.
Worth asking any vendor in this category, us included: which frameworks the score is built on, whether those frameworks' published reliability was established in the setting you are buying for, and whether any automated score has been compared against blinded expert raters.
4. What a privacy or procurement reviewer will find
Every patient in every case is synthetic, written from the clinical literature rather than adapted from a chart. Case development needs no patient record and no de-identification step, so no protected health information enters the platform through a case.
What the platform does handle is learner recordings, transcripts, account data, and institutional data, and those records need protection. Learners speak their side of the conversation aloud, which makes voice data sensitive. Voice collection is consent-gated and learners can request erasure. A recording exists to generate that learner's feedback.
Once a case is published, its case text, its rubric, and its scoring prompts are versioned and locked, so a program can identify the exact materials behind any learner's report and inspect the transcript evidence behind each score.
Nothing crosses organizations. Every project, learner, transcript, and score belongs to one organization, and no role reaches outside it. Inside an organization, roles decide who sees whose results: details are in the roles and permissions guide, and the full data-handling picture is on our trust page.
If your security review needs something these pages don't cover, ask us. We'll tell you what exists and what doesn't.
5. What it takes to run
A browser, and that is the whole list. ClinicalSim runs on any phone, tablet, or desktop with nothing to download and no app to install. A typical encounter takes 3 to 10 minutes, short enough to fit a clinical day and repeat as often as a learner needs. The platform currently supports English.
There are four roles. Members practice and see only their own work. Project Managers build a cohort, assign cases, and see that cohort's results and nothing else. Admins see every program in the organization and control who belongs to it. An Owner is an Admin who can also close the account. Faculty review assigned learners' progress, recordings, transcripts, and feedback without attending a session, and programs get cohort progress views, progress reports, exports, and bulk invitations.
The one thing IT may need to do is allow the voice service. Hospital firewalls, VPNs, and web filters sometimes block it, which used to look to a learner like a slow network. The briefing page now runs a connection test that checks whether the network can actually reach the voice service and, when it can't, says so and names what to ask IT to allow.
6. How an engagement starts
An engagement starts with a communication priority your institution already recognizes. In a communication assessment, one group practices a conversation or two with AI patients against the standard your institution already holds, and the report shows strengths and areas for focused practice in the participants' own words. A longer program can include that baseline, and a continuing engagement extends practice and review across more teams.
There is no price on this page. We scope each engagement with the institution, around the conversation it wants to strengthen and the people who will practice it, and quote it directly.
To see what leaders receive at the end of an assessment, read one partner health system's anonymized cohort report, including the scope rule, its limits, and what a second round of practice looks like.
7. How this fits with what you already run
A standardized patient program
ClinicalSim extends a standardized patient program and does not replace it. SP encounters remain the gold standard for live coaching and high-stakes assessment.
Another AI simulation tool
Ask any vendor, us included, which named standard the score is built on, whether each score quotes the transcript behind it, and what the score must not be used for.
Patient experience surveys
A survey reports how patients experienced communication after the visit. ClinicalSim measures the conversation itself, in simulation: what the participant said, scored against the standard your team already teaches. ClinicalSim does not predict patient experience scores, readmissions, safety events, claims, or other clinical or business outcomes. Any comparison with institution-held outcome data requires a separate study plan.
8. How to see it before deciding
Five complete encounters are published with the audio, the full transcript, and the entire feedback report. No sign-in, no form, no call first.
Each example shows strengths, areas for focused practice, and the participant's own words behind the feedback. Read the complete reports to judge whether the scoring is specific, useful, and appropriate for formative practice.
9. What we will not claim
Scattered across this site are the things we've decided not to say. Collected in one place, they are probably the fastest way to work out whether we're worth your time.
- That our scoring is more accurate or more valid than a faculty member's read. We don't have the validation data to say it, and we won't say it until we do.
- That a framework's published reliability transfers to an AI score. It was established with trained human raters watching real encounters, which is not what happens here.
- That ClinicalSim replaces a standardized patient program. It extends one.
- That we can price your malpractice risk or benchmark your institution against another. We do neither.
- A study outcome before the study's owners have confirmed the record in writing.
- That ClinicalSim holds an approved-course designation with any malpractice carrier. It holds none, and that approval is the carrier's decision rather than ours.
- Any effect on HCAHPS. No study has tested ClinicalSim against HCAHPS scores, and we claim no score-to-survey correlation.
10. Questions buyers actually ask
What should a program consider before adopting an AI communication training platform?
Review the platform's intended use and limits, the evidence behind its claims, data handling, rollout needs, how an engagement is structured, and the experience of the people who will practice. Five complete AI patient simulations are published with audio, transcript, and full reports so each reviewer can evaluate the same facts before a scope or pricing call.
Does ClinicalSim replace a standardized patient program?
No. ClinicalSim extends a standardized patient program rather than replacing it. SP encounters stay the high-stakes assessment and the place live coaching happens, and ClinicalSim adds repeatable practice between them without booking another actor or room. Because it is audio only, it cannot assess eye contact, body language, or physical presence, so those skills stay with live encounters and human review.
Can a ClinicalSim score carry a high-stakes decision on its own?
No, and ClinicalSim does not claim it can. The communication frameworks ClinicalSim applies were built for trained human raters observing real encounters, and that is the setting in which their published reliability was established. Scoring them with AI in a simulated encounter goes beyond that setting, so a framework's reliability does not carry over to a ClinicalSim score. Each score is a formative signal backed by verbatim transcript evidence, meant to sit alongside faculty observation rather than substitute for it.
Does patient data enter ClinicalSim?
No patient data enters through a case. Every patient in every ClinicalSim case is synthetic and written from the clinical literature rather than adapted from a chart, so case development needs no patient record and no de-identification step. The platform does handle learner recordings, transcripts, account data, and institutional data, and those records need protection. Voice collection is consent-gated and learners can request erasure.
What does a program need in place to run ClinicalSim?
A browser and nothing else. ClinicalSim runs on any phone, tablet, or desktop with no download and no app to install, and a typical encounter takes 3 to 10 minutes. The platform currently supports English only. One thing IT may need to do is allow the voice service, because hospital firewalls, VPNs, and web filters sometimes block it; the briefing page runs a connection test that detects this and names what to ask IT to allow.
How does an engagement with ClinicalSim start, and is there a published price?
An engagement usually starts with a communication assessment: one group practices a conversation or two with AI patients against the standard the institution already holds, and the report shows strengths and areas for focused practice in the participants' own words. A longer program can include that baseline, and a continuing engagement extends practice and review across more teams. No price is published. ClinicalSim scopes each engagement with the institution and quotes it directly.
Can I see a real ClinicalSim encounter before talking to sales?
Yes. Five complete AI patient simulations are published at clinicalsim.ai/examples with audio, the full transcript, and the entire feedback report, with no sign-in and no form. Each report shows strengths, areas for focused practice, and the participant's words behind the feedback.