← glacis.io
Confidential December 2025

Response to CoFound
Diligence Materials

To: Jordan Wan (CoFound)
From: Joe Braidwood & Dr. Jennifer Shannon

01 GTM Segments & ICPs 02 Design Partners 03 Revenue Line of Sight 04 Intellectual Property 05 Competitive Landscape 06 The Ask

Executive Summary

GLACIS is building the evidence layer for AI governance — cryptographic attestation infrastructure that turns runtime controls into independently verifiable proof. Our wedge isn't "better guardrails" (that's becoming table stakes). It's evidence-grade enforcement: portable receipts that third parties can verify without trusting our dashboard.

You nailed the core insight in our call: "the buyer doesn't even know how to evaluate" — and vendors have to control the narrative with proactive dossiers. That's exactly what we productize. Healthcare is our beachhead because digital health vendors are getting blocked by hospital security reviews, not because their AI is unsafe, but because they can't prove it's safe in a way procurement teams can verify.

Colorado AI Act — In Your Backyard

The most frontier AI regulation in the US becomes enforceable June 30, 2026. Your state's AG is about to have a lot of leverage. NIST AI RMF / ISO 42001 compliance creates an affirmative defense — we make that provable with cryptographic evidence. Every digital health company selling into Colorado-based health systems will need this.

Early Validation — All Inbound

Three design partners committed (nVoq, Prompt Opinion, PraxisPro). Credo AI's CEO (Forrester Wave leader) reached out directly — meeting at JPM to discuss integration. Mayo Clinic in active prospecting. Zero marketing spend.

01 GTM Segments and ICPs

Primary ICP: Digital Health AI Vendors

We sell to the vendor who is blocked, not the hospital who is busy.

The Insight

Hospital security teams are gatekeepers, but they're not the ones with urgent purchase intent. The pain sits with AI vendors trying to get through security review — losing deals, delaying deployments, burning cycles on manual audit responses.

Target Profile

  • Digital health SaaS deploying AI into clinical workflows
  • Series A-C, $10-100M revenue range
  • Already have guardrails (or think they do)
  • Getting blocked on "prove your AI is safe"

Titles: VP Engineering, Head of Product, CCO, CEO

Why They Buy

  • Reduce time-to-approval with health systems
  • Produce evidence artifacts security reviews accept
  • Avoid building bespoke audit infrastructure
  • Get ahead of regulatory deadlines

Secondary ICP: Health System Security Teams

We're not selling to hospitals directly in Phase 1, but we need them to pull our evidence format into evaluation workflows. When a hospital says "we accept GLACIS attestation packs," vendors have to adopt us.

The play: Partner, don't sell. Get our evidence format accepted as a standard artifact in their security review process. This creates demand pull from vendors.

Tertiary ICP (2026+): Insurance Carriers

Insurers can't price AI risk because they lack actuarial data on what controls actually work. We provide the telemetry and evidence infrastructure that makes AI deployments insurable.

This is Phase 2. The prerequisite is production deployments generating real attestation data.

02 Design Partners & Validation

All Inbound — Zero Marketing Spend

Everything below came from organic discovery — LinkedIn content, HLTH networking, word of mouth. We haven't run a single campaign. The pull is real.

Design Partners (Committed)

Company Status Deal Structure What We'd Prove
nVoq LOA in legal review (2nd round) — signing this week/next 90-day pilot → $25K ARR conversion Evidence pack closes enterprise security review
Prompt Opinion Tech kickoff scheduled January Design partnership AI governance for clinical decisions

Active Prospecting

Company Status What They Need
PraxisPro Design partner, diligence underway Evidence-grade AI governance for enterprise
Mayo Clinic Platform CIO conversation AI governance framework for clinical AI

Risk Reversal Guarantee

We're confident enough to offer: No deployable evidence pack? Full refund. Design partners can pilot with zero downside — if we can't produce verifiable attestation artifacts that pass their customer's security review, they pay nothing.

What We're Proving

Technical

  • Zero-egress eliminates BAA friction
  • <50ms latency overhead
  • Crypto attestation holds under load

Commercial

  • Evidence Pack is what buyers want
  • Verifier-first UX resonates
  • Faster to sell to blocked vendors

Wedge

  • "Evidence-grade" differentiates
  • Third-party verifiability is the gap
  • PHI egress is right control family
If we can produce independently verifiable evidence that a vendor's AI controls executed correctly, we can collapse the security review timeline from months to days — and that's worth $50-100K/year to a digital health company losing deals to procurement friction.

03 Line of Sight to Revenue

$500K–$1M
ARR Target by Dec 2026
10–20
Paying Customers

Two Revenue Tracks

Track A: Evidence SaaS

Price: $50-100K/year per enterprise

Path: 10 customers × $75K = $750K ARR

Motion: Vendor blocked → we deploy → they close deal → they pay us

Track B: Evidence-at-Procurement

Price: $25-50K/year + health system co-investment

Path: 20 vendors × $40K = $800K ARR

Motion: Health system accepts format → vendors come to us

Timeline to Revenue

Q1 2026

Convert & Ship

  • Convert 2-3 design partners to paid pilots ($15-25K each)
  • Ship Evidence Pack v1.0
  • Establish 1-2 health system relationships
  • JPM Healthcare: announce design partners
Q2 2026

Scale Contracts

  • Convert pilots to annual contracts ($50-100K)
  • First health system accepts GLACIS format
  • 5-7 total paying customers
  • Colorado AI Act effective (June 30)
Q3 2026

Regulatory Tailwind

  • Second wave from Colorado compliance
  • EU AI Act main compliance (August 2)
  • Begin insurance carrier conversations
  • Target: 10-15 paying customers
Q4 2026

Series A Positioning

  • $500K-$1M ARR run rate
  • Evidence format = procurement standard (2-3 health systems)
  • Insurance pilot underway

Assumptions & Risks

Key Assumptions

  • Evidence-grade enforcement is the wedge
  • Vendors will pay to unblock procurement
  • Health systems will accept third-party attestation
  • Regulatory deadlines create urgency

Key Risks

  • Design partners don't convert → tight feedback loops
  • Guardrail vendors add evidence → patent moat
  • Health systems build in-house → network effects
  • Sales cycle longer → start with smaller vendors

04 Intellectual Property

The Flight Recorder Concept: Treating AI monitoring as a cryptographically constrained, tamper-evident flight recorder — evidence stays local, only commitments export, and third parties can verify without seeing sensitive data.

70+
Patent Claims Filed
4
Core Patent Families

Filed with Fenwick & West, November 2025

Patent Families

Family A: Integrated Non-Egress Attestation with Co-Epoch Binding

  • Governed canonicalization (CBOR per RFC 8949 with version-pinned encoders)
  • Policy-scoped PRF for deterministic, auditor-reproducible sampling
  • Two-tier attestations: L0 (all requests) and L2 (sampled with full prf_tag)
  • Co-epoch binding to witness-derived binary hash and network state
  • Digest Publication Ledger (DPL) enabling third-party verification without evidence access

Family B: Cryptographically Gated, Self-Stabilizing Control

  • Metrics aggregated only from cryptographically verified receipts
  • Controller with bounded invariants (sampling_prob, queue_max, rate_limit)
  • ControlActions require five gates: signature validity, epoch currency, parameter bounds, co-epoch receipt, co-epoch NETATT

Family C: Insurance Risk Pricing and Parametric Triggers

  • Parametric trigger conditions verified from attestation data
  • Automated payout execution without manual claims adjustment
  • Zero-knowledge risk assessment without content custody
  • Syndicated coverage via shared attestation format

Family D: Statistical Safety Signal Protocol (S3P)

  • Per-epoch secret nonce generated by witness CSPRNG
  • Cryptographic random sampling at configurable rates (0.1-1%)
  • Post-epoch nonce publication enables exact sampled-set reconstruction
  • Clopper-Pearson confidence intervals for actuarial pricing

Why This Matters

The patents protect the verification layer, not just the enforcement layer. Competitors can build guardrails. They can build monitoring. But producing cryptographic evidence that a third party can verify without trusting the vendor dashboard — and turning that into insurable, parametric trigger events — that's our protected territory.

Market Expansion: Beyond Healthcare

Financial Services

Model risk management attestation for regulators

Insurance/InsurTech

AI underwriting controls with parametric triggers

Autonomous Systems

Safety envelope verification for robotics and vehicles

Enterprise AI

Procurement compliance for Fortune 500 AI adoption

The patent portfolio explicitly covers insurance integration, positioning us for the $50B+ AI liability insurance market as it emerges.

05 Competitive Landscape

Market Structure

The AI governance market segments into three layers:

Layer 1

Runtime Security
Pre-inference filters, prompt injection defense

Layer 2

Monitoring
What happened, post-hoc analysis

Layer 3 ← GLACIS

Evidence/Attestation
Verifiable proof for third parties

Most vendors play in layers 1-2. Layer 3 is the gap.

Competitive Matrix

Vendor
What They Do
What They Don't Do
Lakera
Catch threats in real-time
Prove to third parties they caught them
Alinia
Enforce policies
Produce verifiable evidence for external audit
WhyLabs
Monitor model behavior
Anything that generates revenue (Apple acqui-hire)
Credo AI
Manage governance programs
Runtime attestation, zero-egress deployment
Onboard AI
Standardize buyer evaluation
Produce vendor-side evidence artifacts
EQTY Labs
Hardware-rooted attestation
Edge deployment, commodity infrastructure
GLACIS
Evidence-grade enforcement

Our Sharp Positioning

"We turn runtime controls into independently verifiable evidence that closes audits and procurement — portable receipts a third party can check without trusting the vendor dashboard."

What We Say

Third-party verifiable evidence. Closes procurement. Security teams verify in 2 minutes.

What We Don't Say

"Better guardrails" (feature war) • "More monitoring" (WhyLabs lesson) • "Governance platform" (Credo's game)

Integration Strategy

Rather than competing head-on with guardrail vendors, we position as the evidence layer that plugs into the ecosystem:

This makes us infrastructure, not just another vendor in the stack.

06 The Ask

Milestones

  • 10 paying customers
  • $500K-$1M ARR
  • Evidence format accepted by 2-3 health systems
  • GLACIS Attestation Profile 1.0 (open spec)

Use of Funds

  • 60% Engineering: Evidence Pack, verifier UX, integrations
  • 30% GTM: Convert design partners, health system relationships
  • 10% Operations: SOC 2, legal, infrastructure

Timeline to Series A: 18-24 months with demonstrated revenue and evidence format adoption.

Why CoFound

Healthcare DNA

Your Oscar and ZocDoc background means you've lived the buyer-side pain. Three healthcare unicorns in portfolio. You know which digital health companies are hitting procurement walls right now — and you mentioned two CTOs you'd intro us to.

Insurance Thesis

You immediately saw the reg tech / insurance enabler angle. Our patents explicitly cover parametric triggers for AI liability insurance. We're applying to Lloyd's Lab. This is infrastructure for the $50B+ AI insurance market as it emerges.

Portfolio Fit

Your Seattle companies (including Friday Harbor) and broader healthcare portfolio are exactly the ICP we're targeting. We'd love to explore whether any are facing the procurement friction we solve — and whether there's a path to design partnership or pilot.