Unlocking Velocity
for Regulated Startups
AI2 INCUBATOR · CLOUDFLARE LAUNCHPAD
The Ecosystem Friction
The “Pilot Purgatory” Problem
The Engine is Ready
Gemini 3.0 Pro and Vertex AI are the most powerful engines on the market. But your highest-value startups (Health, Fintech, Insurance) are driving with the handbrake on.
The Liability Gap
It's not a capability gap; it's a liability gap. Founders can build a demo in a weekend, but it takes months to prove "safety" to an enterprise CISO.
Core insight: The problem isn't missing guardrails. It's missing proof that guardrails executed.
Average time to production for regulated AI workloads due to audit requirements.
The Timing
The Regulatory Cliff is Here
Startups planning for 2026 deployment must solve for these dates today.
Colorado AI Act
Requires documented impact assessments and verifiable consumer protection.
EU AI Act
Full enforcement. Article 12 mandates "automatic recording of events" for high-risk AI.
California ADMT
Explicit coverage for healthcare decision support. Pre-deployment audit requirements.
The Solution
The “Digital Flight Recorder” for Vertex
Cryptographic Chain of Custody
We don't just filter bad prompts. We issue a cryptographic receipt for every single interaction with Gemini. It turns "subjective safety" into "objective evidence."
Compliance in a Box
Glacis is the native "Compliance Layer" for the modern AI stack. We allow startups to hand a mathematical proof to their auditors instead of a CSV file.
"id": "evt_89234...",
"timestamp": "2025-12-05T10:22:01Z",
"policy": "HIPAA_STRICT_V2",
"verdict": "SAFE",
"checks": [
{"check": "PII_REDACTION", "status": "PASS"},
{"check": "TOXICITY", "status": "PASS"}
],
"signature": "ed25519:7f8a92..."
}
Architecture
Zero-Egress Attestation
Glacis runs as a sidecar in your GKE cluster, screening both request and response.
App / GKE Pod
GLACIS Sidecar
PII Redaction
Policy Check
Receipt Sign
Vertex AI
Gemini API
Glacis Witness Service
Attestation anchoring & audit export
Zero Egress
PHI/PII redacted before leaving your VPC. Only sanitized data reaches Vertex.
Bi-directional
Screen both prompts and completions. Catch PHI leakage in model responses.
GKE Native
Deploy via Helm. Runs as sidecar in your existing GKE workloads. No network changes.
Design Partner
nVoq: Home Healthcare Voice AI
Target Pilot
Ambient Scribe for Home Health
Voice AI that documents patient visits in real-time. Currently on AWS, scoping migration to Vertex AI.
The blocker: Their compliance team needs 6+ months to validate any new AI provider. Glacis cuts that to days.
Why Google Wins
Glacis makes switching to Vertex AI frictionless for compliance teams. Same audit trail, new provider.
The Ask
Intro to 3-5 healthcare startups in Google for Startups portfolio who are stuck on compliance.
The Proof Point
If we unblock them in 30 days, let's talk about making Glacis the default trust layer for the portfolio.
The Moat
Insurance is the Unlock
"Can you get AI liability coverage?"
Today, insurance carriers can't price AI risk because there's no evidence trail. Glacis creates the auditable record that makes AI insurable.
Actuarial Data
Every attestation creates pricing data. Carriers can model risk based on actual policy execution, not assumptions.
Claims Defense
Cryptographic proof of what happened. When litigation hits, you have evidence—not reconstructed logs.
Premium Reduction
Early adopters are seeing 30-50% lower premiums. Better evidence = better rates.
The moat: We become the trusted third party that both cloud providers and insurance carriers rely on.
Team & Pedigree
Built by People Who Lived It
Joe Braidwood
CEO
Scaled SwiftKey from 0 to 1B devices.
SwiftKey: Founding team. Sold to Microsoft for $250M.
Vektor Medical: CSO of FDA-cleared cardiac AI. Lived the regulatory nightmare.
Dr. Jennifer Shannon
MD, Chief Clinical
FDA Authorization for AI.
Cognoa: Medical Director who helped secure the first FDA authorization for an AI diagnostic (Autism).
BACKED BY
AI2 Incubator · Cloudflare Launchpad
Next Steps
Let's Explore Together
We'd love an introduction to a few healthcare startups in your portfolio who might benefit.
If we can help them, let's talk further.