01 / The Vendor
Your use case is not on the roadmap.
The vendor will get to it in two quarters. You needed it last week.
100+
healthcare skills across 20+ clinical categories
72K+
ICD-10 codes with HCC V24/V28 fluency
265+
payer policy questionnaires, refreshed continuously
30+
healthcare-specific API endpoints, growing
The Pattern
Healthcare teams already know what needs to be fixed. The platforms underneath them never caught up. Until now, the only path forward was to build the workaround yourself.
01 / The Vendor
Your use case is not on the roadmap.
The vendor will get to it in two quarters. You needed it last week.
02 / IT
Your build request is in the backlog.
It has been there since Q2. The people who would build it are wired up to four other priorities.
03 / Generic Builders
They do not speak healthcare.
No HCC. No payer policy. No clinical taxonomy. You would have to build all of that context from zero.
04 / The Workaround
So you built it yourself.
It works. It is also fragile, undocumented, and only you know how it runs.
Clinicians Build on Gwen
Nurses, physicians, clinical informaticists, CDI specialists, medical coders. The professionals who know exactly where every workflow breaks. Gwen hands them infrastructure that respects their judgment and then gets out of the way.
UM Nurse · Prior auth pre-screen
An Ai Worker pulls the payer policy, matches against the queue, surfaces the three clinical anchors that matter, and routes the edge cases for a second look. Every recommendation stays under nurse review.
CDI Specialist · Physician query helper
The Ai Worker reads the chart, flags conditions that warrant clarification, and drafts the query in the specialist's signature style. Review. Send. The afternoon stops disappearing into queries.
Oncology Informaticist · Tumor board prep
Labs, imaging, prior treatments, complications, stitched into the format the tumor board actually runs on. The hour before the meeting belongs to clinical thinking, not paginating.
The art of medicine stays with the clinician
The Ai Worker handles the time-burdensome work. Every recommendation explains itself. Every output carries an audit trail. The clinician stays in the loop where judgment lives.
Two Doors, One Engine
Same versioned skills. Same audit trail. Same evidence on every output. Your platform team and your subject matter experts finally build on the same substrate.
Door 01 · Engineers
Sync and async clients. Auto-retry. Private VPC. Typed Pydantic results. Pin to a specific skill version for drift-free production.
# Pull a prior auth recommendation
from penguinai import Gwen
gwen = Gwen(version="v7")
result = gwen.skills.prior_auth_review(
member_id="M-21044",
service_codes=["99213", "93000"],
policy="BCBS-FL-2026",
)
print(result.recommendation, result.evidence)
Door 02 · Operators
Marketplace, Studio, Agent Builder, Cognitive Services. The same versioned skills, exposed through a no-code surface for the people closest to the work.
Describe the workflow
"When a prior auth lands, pull the policy from the payer, match the service codes, and route to a nurse reviewer if confidence drops below 85."
Resulting skill chips
Same hardened skills. Same audit trail. Same humans in the loop.
One library · Four surfaces
What's Built In
Four foundations under every Ai Worker you build with Gwen. Nothing to assemble. Nothing to license separately. All of it live the first time you log in.
01
Proof
72,000+ ICD-10 codes
HCC V24/V28
265+ payer policies
Quarterly refresh
02
Proof
Versioned skills
Public eval leaderboards
Pydantic-validated outputs
03
Proof
20+ enterprise connectors
FHIR R4
X12 · NCPDP
Browser automation
04
Proof
Glassbox reasoning
HIPAA audit log
Source-span evidence
Human in the loop
How It Works
Four steps. Same flow whether you ship through the SDK or through Studio.
01
Pick or build
Marketplace covers the common workflows. For everything else, plain English maps to the right skills automatically.
02
Map your inputs
Gwen handles the parsing, OCR, and routing. You confirm what good output looks like.
03
Run with humans
Confidence gating routes the edge cases to your team. Your reviewers approve, modify, or reroute. Nothing gets shipped without a human in the loop where it matters.
04
Ship & audit
Full HIPAA audit log on every call. Evidence traces back to source spans. Iterate without disrupting what is already live.
Pre-Built Ai Workers
A sample of what is live in the Marketplace today. Each one is a starting point. Fork it, extend it, version it, ship it.
Utilization Management
Pulls payer policy, matches against service codes, surfaces clinical evidence, routes for human review when confidence drops.
v3.2 · 91% accuracy
Revenue Cycle
V24 and V28 fluent. Reviews charts, surfaces missed conditions, attaches supporting documentation for coder review.
v5.1 · 94% accuracy
Claims & Adjudication
Routes incoming claims by policy, payer, and complexity. Flags denial risk before submission.
v2.8 · 89% accuracy
Appeals
Pulls denial reason, matches to coverage policy, drafts evidence-backed appeal letter for human signoff.
v4.0 · 92% accuracy
Clinical Operations
Reduces 90-minute chart reviews to under 5. Surfaces relevant clinical history, medications, and prior episodes.
v6.3 · 93% accuracy
Member Operations
Real-time eligibility checks across payer portals. Browser automation where the API stops.
v3.7 · 96% accuracy
GWEN
▸ Live now at gwen.penguinai.co
$10 in credits the moment you sign up. No credit card. No sales call. No commitment until the product earns it. The people closest to the problem can finally fix it themselves.
Trusted by clinical informaticists, revenue cycle leads, and platform engineers at the largest payers and providers in the country.
Healthcare administration at the speed of patient need.