VANCOUVER, British Columbia and FAIRFIELD, Conn., Oct. 20, 2026 (GLOBE NEWSWIRE) -- CharliAI and Protegrity today announced a partnership that lets regulated enterprises run AI agents on their most sensitive data by controlling what each agent can see and do. Protegrity protects the sensitive data agents use and the knowledge they could piece together from it. CharliAI governs what agents are allowed to do and keeps a forensic record of what they did.
AI agents are moving out of pilots and into production workflows where they access critical data, change records and trigger actions in other systems. Security and audit teams must answer for every step, and most aren’t ready to sign off. In EMA’s 2026 State of AI Friction research, 83% of IT and security leaders report that security or compliance reviews have delayed AI from reaching production. Meanwhile, 69% already run agents that write to databases and trigger external APIs, yet still rely on controls built for humans querying data, not agents acting on it.
“Organizations aren’t short on AI ambition. They’re short on a way through the risk review,” said Christopher Steffen, VP of Research at Enterprise Management Associates (EMA). “Slow down for the review and you pay in delayed, diminished AI. Speed up and you pay in exposure. That’s the AI friction tax, and most organizations are paying both sides of it. The way out isn’t a faster review. It’s protection built into the workflow, so agents can be cleared to do the work they were built for.”
Getting an agent through review comes down to two questions: what is it allowed to see, and what is it allowed to do? Agents reason, hand work to other agents and act at machine speed, so a one-time access check can’t answer either question, nor can a dashboard that reports exposure after the fact. Answering both questions, while agents reason and act, takes data protection and runtime control working together.
“The market is finally waking up from the AI hype cycle and realizing that capability is not clearance,” said Kevin Collins, CEO of CharliAI. “Just because an agent can execute a complex workflow doesn’t mean it has the authority to do so unsupervised. Enterprises want the productivity gains AI promises, but they cannot afford to hand over the keys to their data and systems to get it. We give them the runtime control and forensic proof they need to move autonomous agents into production.”
The integration embeds protection into the agent’s workflow rather than at the perimeter.
Protegrity protects sensitive data before any model, agent or tool receives it. Agents work with semantically protected values that keep their meaning but not their risk, and policy decides what data to release based on who is asking and why.
CharliAI checks each agent’s identity and authority before it acts and stops execution when behavior goes out of bounds. Decisions made and actions performed are recorded as the work happens, giving security and audit teams evidence of what the agent was shown and what it did.
“The agent that should worry you isn’t just the one that goes rogue. It’s the one that does exactly what it’s allowed to do and works out things nobody meant it to know,” said Michael Howard, CEO of Protegrity. “Every agent is a new insider, one that never signed an NDA. Protecting your data and your knowledge means when an agent does something unexpected, it’s an anomaly, not an exposure. Pair that with control over what agents do, and you get knowledge arbitrage: the most value from your data, with the least exposure.”
This partnership allows banks, capital markets firms and insurers to buy the integration from CharliAI, which resells Protegrity’s data and AI security solutions with CharliAI’s Ancaeus platform.
About CharliAI
CharliAI provides the forensic control layer for enterprise AI. Its Ancaeus™ platform enables organizations to profile, trace, control and prove AI and agent activity across sensitive and regulated workflows. CharliAI helps enterprises move autonomous AI into production while maintaining control over data, authority and accountability.
About Protegrity
Protegrity secures data, agents and inference for enterprise AI, enforcing one policy across all three layers so sensitive data stays usable, agents stay bounded and restricted knowledge stays protected. Enterprises use Protegrity to move more AI into production, on more of their data, with security built into each step AI takes.
