Clinical decision support you can audit
In Omni, rules decide only what software should decide: whether an intake meets your program's criteria, which clinicians may see a case, and whether a prescription passes your safeguards. Clinicians make every clinical decision. Recommendations show the reasoning behind them, and decisions are recorded with the rule version that made them.
- Eligibility rules, checked in a fixed orderContraindications, state, BMI, consents, mental-health screenClinician review
- Offered only to clinicians licensed and authorized in IllinoisCredentialing verified, license activePassed
- Protocol recommendation, with its decision pathBMI 30 or higher, no contraindications, asynchronous reviewAdvisory
- Lab gate: confirmed bloodwork within 180 daysLow-confidence OCR values never countPassed
- Controlled-substance policy for the drug and stateBlocked, not waved through, if the check is unavailablePassed
Rules decide. Clinicians decide. AI only drafts.
Clinical software earns trust by being predictable. These principles are built into how Omni works, not just how we describe it.
No model makes a clinical decision
Eligibility, case access and prescribing safeguards run on deterministic rules that can be read and audited. The same inputs always produce the same result, and no machine-learning model makes or scores those decisions.
Clinicians make every clinical decision
Decision lanes, fit scores and protocol recommendations are advice. Even a case that clears every rule only becomes eligible for a clinician's review; nothing is ever prescribed automatically.
Eligibility rules change only with two sign-offs
A change to your eligibility rules moves from draft to a compliance review to a medical director's approval before it publishes. Nobody can approve their own change, and every version is kept.
When in doubt, it holds
If no rule matches, the decision service holds rather than allows. If it can't be reached, controlled-substance prescribing is blocked. Lab values read with low confidence don't count until a clinician confirms them.
How a decision flows
From the first intake answer to a signed prescription, each step adds evidence the clinician can see.
- IntakeAnswers are checked against your eligibility rules in a fixed order. Hard contraindications stop the intake before it reaches a prescriber.
- AccessThe case is visible only to clinicians with verified credentials who are licensed and authorized in the patient's state.
- ProtocolThe program's protocol map recommends a path, such as asynchronous review or a video visit, and shows how it got there.
- ReviewThe clinician sees the recommendation, safety flags, care gaps and the patient's health snapshot, then decides.
- SignThe safeguards your brand switches on run at signing. A failed check blocks the signature and states the reason.
Protocols your clinical leads can read
Protocol maps are decision trees built on a visual canvas from yes-or-no checks and "at least M of N" checks over intake answers, BMI, age, state, labs and custom fields. Drafts can be tested on a patient and are never applied to real patients until a map is activated. One map is active per program.
- Readable by clinicians, with no code to review
- Tested on a patient before it goes live
- Clinicians see the full decision path behind each protocol recommendation
What's in the decision-support toolkit
Each capability is labeled with where it stands today: working in Omni's demo environment, in development, or planned.
Governed eligibility rules Working
A written, versioned rule set checked in a fixed order: contraindications, state coverage, age and BMI minimums, consents, transfers of care and a mental-health screen. Weight-loss is the first program configured this way.
Self-harm screening Working
A weight-loss patient who screens positive for self-harm risk on the PHQ-9 always goes to clinical review and is never fast-tracked.
Prescribing safeguards Working
Checked at signing by a separate decision service: controlled-substance rules by DEA schedule and state, recent-lab requirements by program, and in-state license and DEA checks, each switched on per brand.
Credential-scoped case access Working
Only clinicians with verified credentials and an active license and authorization in the patient's state can see or claim a case. Returning patients go back to their own clinician.
Self-expiring credentials Working
Every minute, Omni checks each verified clinician's licenses, DEA and state registrations and malpractice cover, and suspends new cases the moment one lapses.
Protocol Mapper Working
Visual decision trees with yes-or-no and M-of-N checks, tested on a patient before activation, with the full decision path shown to the clinician.
Safety flags and care gaps Working
Flags at the top of each case for critical and abnormal labs, allergies, reported contraindications and medicines such as insulin or sulfonylureas, plus rule-based care gaps such as overdue labs and titration check-ins.
Document import with provenance Working
The Data Hub reads PDFs, scans, photos, Word files and C-CDA documents and proposes labs, medications, conditions, allergies and vitals. Each value shows its confidence, page and exact quote, and your team accepts, edits or rejects it before it reaches the chart.
Patient lab uploads Working
Patients can upload a photo or scan of a lab report. Values read with low confidence are flagged and don't count toward lab requirements until a clinician confirms them.
Decision records Working
Safeguard decisions are written to the write-once audit log with their policy and version, and routing decisions are stored permanently with their reasons and rule version.
Clinical automations Working
Event-driven workflows (intake completed, lab received, refill requested) that wait days or weeks for real events, call the same decision engines, and start from a library of 14 treatment plans.
Governed AI drafting In development
The governance layer is built: drafts wait for a clinician to accept, modify or reject them, and nothing is signed automatically. It isn't connected to a model yet.
AI-assisted document extraction In development
An optional extraction engine using Anthropic's Claude on Amazon Bedrock is built and stays off until it can run under a business associate agreement.
Visit transcription Planned
Transcription of video visits stored with the encounter, so the clinician starts from an accurate record of what was said.
Device and wearable data Planned
Connected scales, blood-pressure cuffs and wearables feeding the health snapshot clinicians already review.
See it in the product
Screens from Omni's demo environment, shown with synthetic data. Select any screen to zoom in.
Where AI fits, and where it never will
No generative AI runs in Omni's clinical workflows today. Every decision on this page is made by deterministic rules and by clinicians.
When AI drafting is switched on, it will be an assistant and nothing more: every draft waits for a clinician to accept, modify or reject it, nothing is signed automatically, and each request is recorded with the model, a fingerprint of the prompt, the output and the clinician's response. Model calls are designed to run on Amazon Bedrock inside Omni's own AWS environment, and each brand can turn AI off entirely.
- Draft a note or a visit summaryThe clinician accepts, edits or rejects itAllowed
- Propose values extracted from a documentA person reviews each value before it's chartedAllowed
- Decide whether a patient is eligibleDeterministic rules onlyNever
- Approve, sign or route a prescriptionClinicians and safeguards onlyNever
Built with regulators' expectations in mind
Omni is decision support for licensed clinicians. The design follows the principles regulators use to judge clinical software.
FDA clinical decision support
Omni is designed around the criteria for non-device clinical decision support in the 21st Century Cures Act: its recommendations go to licensed clinicians, the basis for each one is shown so they can review it independently, and it doesn't analyze medical images or device signals.
Transparency of decision support
In the spirit of ONC's HTI-1 transparency rules, decisions record the rule, version and reasons behind them, protocol recommendations show their decision path, and every eligibility-rule change has a recorded author, reviewer and approver.
HIPAA
Decision support runs inside each brand's isolated environment on HIPAA-eligible AWS services. The audit log records each decision's outcome and policy version without copying patient details into the log.
Omni has not been cleared or approved by the FDA and is not ONC-certified health IT. Program eligibility criteria are set by each practice's clinical leadership, and how a regulation applies depends on how each practice uses the software, so confirm the position for your practice with your own counsel.
Questions medical directors ask
Does Omni use AI to approve or decline patients?
No. Eligibility is decided by deterministic rules set by your clinical leadership, and a clinician reviews every case before anything is prescribed. No generative AI runs in Omni's clinical workflows today.
Who can change the rules?
Eligibility rules change through a governed draft: a compliance review, then sign-off by a medical director who didn't write the draft. The new version applies from the next intake, and every previous version is kept. Protocol maps are edited by your clinical leads and can be tested on a patient before they're activated. Prescribing safeguards are brand settings, and every change is logged with its before and after values.
What happens when a check can't be completed?
Omni holds rather than guesses. If no rule matches, the decision service holds. If it can't be reached, controlled-substance prescribing is blocked. Lab values read with low confidence don't count until a clinician confirms them, and a page that fails to load says so instead of showing an empty list.
When will AI drafting be available?
The governance layer is built. Connecting it to a model is in development, on Amazon Bedrock inside Omni's AWS environment, and it will be switched on only once it can run under a business associate agreement. Each brand will be able to keep it off.
Can we see why a patient was routed a certain way?
Yes. Routing decisions are stored permanently with their reasons and rule version, clinicians see each protocol recommendation's decision path, and administrators get routing analytics by outcome, lane and program.
See a decision trace on your own protocol
Bring your eligibility criteria and prescribing policies. We'll configure them in Omni and walk your medical director through exactly how each decision is made and recorded.