Gartner's own forecast of 150,000 AI agents per Fortune 500 company by 2028 makes agent governance a real, near-term problem, not a hypothetical one, even though Astralis itself is an early, unverified vendor claim.
- Data privacy engineering company Ethyca launched Astralis today, a platform that governs how enterprise AI agents use company data in real time.
- Astralis combines an automated privacy assessment tool that cuts review time from 40 to 60 hours down to 20 to 40 minutes, with a Purpose-Based Access Control engine that screens each data request as it arrives.
- Gartner forecasts the average Fortune 500 company will run more than 150,000 AI agents by 2028, up from fewer than 15 last year.
- One large U.S. financial institution already uses Astralis to automate 6,000 data access requests per second.
- Ethyca’s existing customers include The New York Times, Ramp, and Advance Magazine Publishers, Condé Nast’s parent company.
What Happened
Data privacy engineering company Ethyca launched Astralis today, a platform that governs how enterprise AI models and agents use company data in real time.
Astralis combines two parts. Its Large Language Regulatory Model automates privacy and AI assessments, cutting a process that previously took 40 to 60 hours down to 20 to 40 minutes.
A separate Purpose-Based Access Control engine screens each data request as it arrives, tying the approval to the specific reason behind it. The entire platform runs inside a customer’s own cloud environment, so sensitive data never leaves the organization’s control.
Ethyca said one large U.S. financial institution already uses Astralis to automate 6,000 data access requests per second, or hundreds of millions of governed decisions each month. Existing Ethyca customers include The New York Times, Ramp, and Advance Magazine Publishers, Condé Nast’s parent company.
Why It Matters
Gartner forecasts the average Fortune 500 company will run more than 150,000 AI agents by 2028, up from fewer than 15 last year, and most compliance teams have no answer for why a given agent reached a given dataset. For Ops and Engineering teams scaling AI agent deployment, Astralis targets a real, growing gap between deployment speed and governance capacity, not a hypothetical future problem.
Ethyca’s efficiency numbers, the 40-to-60-hour assessment cut to 20 to 40 minutes and the 6,000-requests-per-second figure, are the company’s own claims, with no independent audit cited. Ethyca is also a small, single vendor competing in a governance category that larger platforms and cloud providers are likely to enter directly as agent sprawl becomes a bigger problem.
Organizations are not ready for the AI onslaught. Cillian Kieran, Founder and CEO, Ethyca
Bottom Line
Watch whether Ethyca’s named enterprise customers publicly confirm results at the scale claimed, and whether major cloud providers or identity platforms launch competing agent-governance tools as Gartner’s agent-sprawl forecast plays out. This category is likely to get crowded fast given how central the problem is becoming.
For ops and compliance teams scaling AI agent deployment, per Relve, an AI tools intelligence platform, the practical question is whether a governance tool ties access directly to real-time data requests, the way Astralis claims to, rather than relying on periodic reviews that go stale the moment business needs change.
