Glacis: Creating Trust in AI Through Tamper-Proof Verification
In a digital landscape increasingly shaped by artificial intelligence, the need for accountability and trust has never been more crucial. As regulatory tensions emerge between federal and state governments over AI oversight, Seattle-based startup Glacis is positioning itself at the intersection of innovation and responsibility, offering a solution that could become fundamental infrastructure for the future of enterprise AI deployment.
Joe Braidwood and Dr. Jennifer Shannon founded Glacis with a simple yet powerful concept: create tamper-proof “receipts” for every AI decision, allowing companies to demonstrate that their safety systems operated as intended. The startup, backed by the AI2 Incubator, essentially functions as a “flight recorder for enterprise AI,” generating immutable records of AI operations. This innovation comes at a pivotal moment, as reports suggest the White House is preparing an executive order that would direct federal agencies to challenge state-level AI regulations. In this environment, Glacis transforms from an emerging startup into what Braidwood describes as an “infrastructure necessity” – a neutral, platform-agnostic trust layer in an increasingly complex regulatory landscape where federal authorities might soon actively oppose state AI restrictions.
The company’s founding story reflects the very regulatory complexities it now aims to address. Braidwood, an experienced tech marketing leader, previously founded Yara, an AI-driven mental wellness startup that he ultimately shuttered after just one year of operation. The decision came after confronting Illinois regulations that made AI therapy “effectively uninsurable” and recognizing the potential dangers of AI when interacting with vulnerable individuals experiencing trauma or suicidal ideation. In a LinkedIn post that garnered significant attention, Braidwood explained this difficult decision and open-sourced the safety prompts he had developed, acknowledging that his experience demonstrated “where the boundaries need to be” and highlighted the immense challenges startups face when operating in high-risk AI categories with uncertain liability and regulatory pressures.
This candid admission sparked conversations across industries. Regulators, clinicians, engineers, founders, and insurance executives all reached out to Braidwood, many identifying the same critical gap: the inability to independently verify whether AI safety policies had actually been implemented in practice. This insight became the foundation for Glacis. The technical solution they developed is both elegant and efficient – creating a signed record every time an AI model responds to a query or takes an action. These records capture the input, document which safety checks were executed, and register the final decision. Crucially, these receipts cannot be altered after generation and add minimal processing overhead – less than 50 milliseconds per interaction. This allows regulators and insurers to verify compliance without accessing sensitive personal data, potentially making previously uninsurable AI applications viable by providing proof that systems followed established rules.
The leadership behind Glacis brings complementary expertise to address both technical and ethical dimensions of AI implementation. Braidwood, serving as CEO, brings extensive experience from roles as chief strategy officer at Vektor Medical, co-founder of social TV platform Scener, and chief marketing officer at SwiftKey. His business and strategic background is balanced by co-founder Dr. Jennifer Shannon’s nearly two decades of experience as a psychiatrist. Shannon, an adjunct professor at the University of Washington and former medical director at Cognoa, also serves on the AI Resource Committee for the American Academy of Child and Adolescent Psychiatry, bringing crucial healthcare expertise to a company targeting sensitive use cases.
While still in its early stages, Glacis is already gaining traction in several critical sectors. The company is currently operating in private beta with digital health customers, including nVoq, and strategically focusing on healthcare, financial technology, and insurance – industries where AI decisions can have significant consequences and where demonstrable safety measures are essential. Their participation in Cloudflare’s Launchpad program further positions them within the technical infrastructure ecosystem necessary for secure, verified AI operations. As AI continues to transform industries and regulatory frameworks evolve in response, Glacis represents an emerging approach to reconciling innovation with accountability – creating the transparent, verifiable infrastructure that may allow AI to advance responsibly in highly regulated environments where trust is paramount and the consequences of failure can be profound.
