About

They Check Messages. Project AIR Proves Missions

LOS ANGELES, CA — Vindicara, Inc. today announced the general availability of Project AIR, an open-source runtime security platform that gives AI agents a tamper-evident flight recorder. The MIT-licensed SDK and CLI let developers record, verify, and audit every agent action — from LLM calls to tool use — with a single install: pip install projectair.

Cryptographically signed, tamper-evident evidence for AI agent activityAs AI agents move from demos into production, enterprises are facing a new governance problem: existing tools can inspect messages, logs, and API calls, but they cannot reliably prove what an autonomous agent actually did, what data it touched, which tools it invoked, whether policy was enforced, or who authorized sensitive actions.

Project AIR is built to close that gap.

“Most AI safety tooling checks messages. AIR proves missions,” said Kevin Minn, founder and CEO of Vindicara. “If an AI agent touches patient data, processes a financial workflow, or triggers a sensitive system action, a company should not be left saying, ‘we think this is what happened.’ They should be able to prove it.”

Project AIR is a LangChain-native forensic SDK. With one install, developers can turn agent actions — prompts, tool calls, data access, handoffs, policy decisions, containment events, and human approvals — into a chained, signed, queryable record. These records are designed to support audit defense, legal discovery, incident response, AI E&O underwriting, GDPR DSAR review, HIPAA accounting workflows, and broader governance for regulated AI systems.

The open-source core is MIT licensed and available on PyPI.

Developers can install Project AIR with:
pipx install projectair
air demoAt the center of Project AIR is AgDR, Vindicara’s Agent Decision Record protocol. AgDR creates forward-chained records using modern cryptographic primitives, including BLAKE3 hashing, Merkle commitment, Ed25519 signatures, post-quantum ML-DSA-65 support, and RFC 8785 JSON canonicalization. The result is a forensic chain that can be verified independently, including offline, without relying on Vindicara’s servers.

Project AIR also includes containment workflows for sensitive actions. When an agent attempts an action that requires human approval, AIR can halt execution, require authenticated approval, and write that approval into the forensic record. This creates evidence of who authorized what, when, and under which agent action.

For regulated sectors, this distinction matters. Healthcare, financial services, insurance, legal, and enterprise security teams are increasingly deploying AI agents into workflows where conventional logs are not enough. They need evidence that can survive audit, subpoena, regulatory review, and post-incident reconstruction.

“AI agents are becoming operational actors,” Minn said. “They do not just generate text. They call tools, access systems, transform data, and make policy-relevant decisions. Governance has to move from passive logging to cryptographic proof.”

Project AIR’s initial open-source release focuses on verifiable agent action records, local demos, SDK instrumentation, CLI workflows, and forensic inspection. Vindicara is also developing a commercial Pro layer focused on audit packs, policy controls, governance exports, OpenLineage support, DSAR workflows, and enterprise-grade evidence management.

The company is currently seeking design partners in healthcare, financial services, insurance, and other security-sensitive environments where AI agents touch regulated data.

Project AIR is available now at:

PyPI: pipx install projectair
Demo: air demo
Company: Vindicara Inc.

About Vindicara

Vindicara builds governance infrastructure for AI agents. Its first product, Project AIR, is an open-source SDK that creates cryptographically signed forensic evidence for agent activity, giving enterprises a way to prove what AI agents did, what data they touched, which policies applied, and who authorized sensitive actions.

Vindicara is a Delaware C-corp founded by Kevin Minn, a solo technical founder building at the intersection of AI infrastructure, cybersecurity, compliance, and regulated enterprise adoption.

Media / Design Partner Contact

Kevin Minn
Founder & CEO, Vindicara Inc.
kevin@vindicara.io