Skip to main content
Home
OverviewGwenOSData LayerContext LayerAi WorkforceGwenCLAF
OverviewPayersProvidersHealthcare ServicesEveryone Else
OverviewIntelligent UMIntelligent RCMHCC Coding & Risk AdjustmentMedical CodingHome Health CodingAi WorkersSnowflake Native Apps
OverviewTeamCareersFAQManifesto
BlogPress Releases
Log in to GwenTalk to us

The trust layer

Healthcare is operating in a cold war.
The way out:
a layer both sides can trust.

Payers and providers spend a trillion dollars a year defending themselves from each other. GwenOS is the trust layer that ends the arms race.

Talk to us

The trust problem

The participants in healthcare do not trust each other.

About 20% of what happens in the system is fraud, waste, and abuse. It happens on both sides. With a trillion dollars at stake, every participant plays defense.

01. The 20% lost.

Fraud, waste, and abuse happening in the system today, on both sides of the transaction.

02. The 20% spent defending.

The administrative defense against the fraud, waste, and abuse that might otherwise happen. Checking, validating, questioning, rejecting, appealing, overturning.

03. The 40% total.

The share of healthcare spend consumed by the problem and the defense against it. A trillion dollars a year protecting a trillion dollars. A losing battle for everyone, and every delay lands on a patient.

The prize

Build the trust layer and 40 cents of every dollar comes back.

Spend on the war becomes spend on care. Whichever direction the system takes, delivery gets more efficient and less costly. The trust layer buys optionality, for all of us.

The GwenOS stack

A shared verification layer replaces the arms race.

Most healthcare technology is built for one side to counter the other. GwenOS spans both: shared policies, transparent reasoning on every recommendation, and outcomes either side can verify. Seven layers make it possible. Most platforms stop near the top. GwenOS keeps going down.

Models

Models

Models

GwenOS routes each task to the model that fits.

Frontier models for breadth. Our own LLMs, SLMs, and sovereign models for the determinations healthcare makes millions of times a day.

Agent Harness

This layer is Gwen.

Skills, endpoints, cognitive services, MCP, routing, and memory, with security and governance built in. Models are necessary but insufficient for healthcare work. The harness is the part that knows what a prior auth is.

Entities & Digital Twins

One record per person

Patient, Member, Provider, Payer, and Claim 360. One resolved record per entity, assembled from every system that holds a fragment. Most healthcare AI fails because the agent does not know who the patient is.

Entities & Digital Twins

Data & Integrations

Nothing moves first.

Connectors, clearinghouses, APIs, and industry networks. GwenOS meets data where it already lives.

Snowflake Native Apps

Capabilities & Knowledge

Built once, used by every workflow

OCR, match and merge, entity resolution, evidence extraction, policy management, intelligent queuing. Beside them sit payer policies, codes, clinical guidelines, and your own history, kept live. The tenth workflow deploys faster than the first because all of this is already here.

Surfaces

Where your people work

  • Gwen StudioBuilds an Ai Worker from a policy document.
  • Gwen CoworkerCarries the repetitive work beside your team.
  • Gwen CodePuts healthcare endpoints in your codebase.
  • Dashboards & AnalyticsShows what every workflow did, and how long it took.
  • End User InterfaceWhere reviewers, coders, and analysts work the queue.

Solutions

Five Solutions, ready to run

  • Intelligent UMReads the chart before the nurse does.
  • Intelligent RCMReviews the claim before it leaves and works the denial when it comes back.
  • Medical CodingSuggests the code and shows the line in the chart that supports it.
  • Payment IntegrityChecks the claim against the contract, the policy, and the record.
  • HCC Risk AdjustmentTies every code to its evidence, before the visit and after.

GwenOS

That is GwenOS.

The layer both sides can trust: one record per person, the payer's policies applied in the open, and a full trail of every action. It plugs into the systems you already have.

How it thinks

Different jobs take different intelligence.

General intelligence is available at an efficient cost, and models play a role in everything GwenOS does. No single model carries healthcare.

01. General models.

Where breadth is the right tool, frontier models are readily available and very useful. GwenOS uses them where they win.

02. Harnessed models.

Most healthcare work needs general intelligence harnessed first: the surrounding mechanisms, business rules, context, and controls that make a model safe to put on real workflows.

03. Specialized models.

For the determinations healthcare makes millions of times a day, Penguin Ai builds its own models. Unique intelligence becomes an asset, and GwenOS routes each task to whichever model fits.

Built in

Healthcare ready from day one.

Human in the loop

Configurable per workflow.

SOC 2 + HIPAA

Aligned on day one.

Snowflake-native

AWS and Azure ready.

Glassbox traceability

On every recommendation.

The trust layer at work

Every request is evaluated in real time.

Take prior auth. The payer's policies for each plan and provider group live in the layer. An agent assembles the request on the provider's behalf, applies the plan-specific criteria for that code, and returns exactly what is missing when something is. The guessing game ends, and so does the queue behind it.

The layer extends into claims

Evidence is checked against each payer's rules.

The prior auth is approved, the care delivered. Now the provider needs to be paid. An agent assembles the evidence and checks it against the payer's configured requirements: the signature, the attestation, the adjacent documentation each payer wants to see. Complete claims are approved. Incomplete ones come back naming exactly what is missing. And because trusted approval separates approval from payment timing, it opens payment arrangements that do not exist today.

What ends the war

Less friction becomes a network advantage.

The trust layer is worth more to each side than the war ever was.

01. Providers.

The submission cost, the review cost, and the resubmission churn come off the desk. Hours a week return to clinical work.

02. Payers.

Provider networks are the most valuable thing a payer builds, because providers determine whether patients get healthier or sicker. The payer that takes the friction out builds the best network.

03. Patients.

All of the checking, validating, debating, and overturning delays care. Every one of us will need care in our lifetime. The trust layer gives that time back to the people the system exists for.

From inside the machine

We are building the Operating System we wished we had.

Our team ran data operations at the largest payer and provider organizations in the country. We built the workarounds. We inherited the spreadsheets. If your problems sound familiar to that paragraph, Penguin Ai will sound familiar from the first demo.

Talk to us

See the trust layer running on your data.

A Penguin Ai specialist will walk through how GwenOS evaluates the workflows your team handles today, and what both sides of the transaction get to see.

Talk to us