One governed path to every AI model.

Miso lets a health system use more AI, from more vendors and model providers, without making each new use case a separate privacy, security, contracting, and engineering project. The system decides once which teams and vendors may use which models, for which purposes and data; Miso enforces it on every request. Runs in your cloud. OpenAI-compatible.

Talk to usHow it works

Every new AI project should not need the same review from scratch.

A clinical team prototypes against OpenAI. A vendor arrives using Anthropic. Another department wants Bedrock. Privacy, Security, and Legal answer versions of the same questions each time, and the projects wait.

Miso turns those answers into routes: who may call which model, with what data, under which agreement. A route is approved once and enforced on every request after that. The next project inherits the decision.

Innovation teams, developers, clinical and product owners

Move AI into production faster

Reuse approved infrastructure in place of reviewing every project from scratch. A developer gets one endpoint and a list of the models they may call; the institution's policy decides what is behind it.

Privacy, Security, Legal, IT

Keep institutional control

Decide once what AI may do, then enforce it across applications and vendors. Fewer independently built AI data paths, fewer credentials in circulation, one place to apply the controls. Agreements become part of the execution path: this BAA covers that provider, so this route exists.

Executives, compliance, audit, governance committees

Know what is actually running

Which AI systems are in use, which models they call, what they are allowed to do, and whether actual activity matches the approval, recorded as it happens rather than assembled afterwards from surveys and tickets.

CEO, CFO, finance, department leaders

Know what AI will cost before the invoice

A budget by application, department, and route, enforced before each request is sent. A forecast of next month's spend while there is still time to change it, with the lever behind each number.

One request, going through.

The developer makes a normal model call. Miso checks the caller, the purpose, the data, the model, and the agreement, then either sends the request or refuses it. Either way it keeps the evidence.

Synthetic examples. Model names illustrative.

Inbox Assistant (pilot)→Miso
Caller
unregistered key
Requested model
GPT-5, OpenAI direct
Agreement
none on file
Result
ROUTE DENIED · logged b7c2…04 · no data sent

For the teams that build

"Give me the endpoint and tell me what models I can call."

That is the whole developer experience. Once a route is approved, the team gets an endpoint and a key and writes ordinary model calls. Which provider is allowed, how PHI is handled, which agreement applies: the institution has already decided, and Miso carries the decision behind the endpoint.

For Privacy and Security

From advice to enforcement.

Today Privacy says "you may use this provider for this purpose under these conditions," and then relies on the application team to do it correctly. With Miso the approval is a technical boundary: if an application is not permitted to send PHI to that model for that purpose, the request does not go. The record shows each request against the approval that covered it.

A boundary clear enough that Privacy and Security can let Innovation move faster.

Innovation wants to move. Privacy and Security need to slow things down enough to understand the risk. Today that is settled project by project, and each application has to be trusted to implement the answer correctly. Miso puts the answer in one shared path, so an approval is reused rather than reconstructed, and Privacy can say yes more often because saying yes no longer depends on every application getting the rules right.

Miso closes the gap between the policy the health system approved and the AI activity actually occurring.

Your AI vendors should not each bring their own AI stack.

Each AI vendor arrives with its own model stack and its own downstream dependencies. Give an approved vendor a Miso endpoint. Their application uses the models and providers your health system has already approved, under your agreements and your controls. They never hold your provider credentials, and the vendor's AI activity shows up in the same record as everyone else's.

Make the vendor inherit your AI controls

vendor.ambient-scribe · encounter-note-draft · purpose: treatmentreceived

Patient Maria Alvarezname removed, DOB 03/14/1961age 65, MRN 44810-2mrn removed, presents with worsening dyspnea on exertion over two weeks. Reports orthopnea; denies chest pain. Started on 05/02/2026~4 months ago on furosemide 40 mg.

Modelazure openai · gpt @2026-07-10 · eastus2pinned for this vendor · BAA in force · zero retention
Vendor oweshuman review before usedisclose AI to patient
Recordhash 91d4…c2 ← prev 7c1e…a94 identifiers handled · values not stored · 412 ms · $0.0031
receivedidentifiers filteredrouted to pinned modelobligations returnedrecorded

Synthetic example.

An AI inventory built from actual use, not another questionnaire.

For the applications and vendors routed through Miso, the inventory updates itself from traffic: who is using which models, for what purpose, under which approvals. The privacy officer sees what needs a decision. A builder sees their own applications, spend, and denials.

Every view in the console

Know what AI will cost before the invoice.

The EHR vendor, the AI vendors, and the model providers are all moving to consumption pricing. Miso checks each request against its budget before sending it, and the record it keeps is enough to forecast next month by department. AI spend

your cloud account
model providers ↗
the only thing outside
miso · policy · filter
postgres · records

Miso runs in your cloud environment.

The gateway and the policy store are deployed within your cloud account, so model credentials, governance configuration, and audit records remain under your organization's administrative control. Nothing leaves through Miso unless a route says it may.

Architecture and security boundary