Palo Alto Networks' talks to buy Clutch Security for $500M show that controlling what AI agents can access is now a core security problem. If you connect an AI agent to your CRM, inbox or accounting, decide exactly which systems and data it can touch before you give it more work.
According to TechCrunch, the offer is $500M in cash plus milestone payouts for a three-year-old startup that was valued at $85M in January 2025. Clutch's revenue is approaching $10M a year, so the price is roughly 50 times that. Competitors are being bought too: Cyera bought Oasis Security for $1B, and Cisco bought Astrix for about $400M.
What this means for a small business
You won't buy Clutch. It's built for enterprises with hundreds of agents and service accounts. The underlying problem still applies to a single Claude agent that handles leads or invoices. The common mistake is giving the agent an admin token "so it just works." A lead-processing agent needs read access to the CRM and write access to one table. It needs nothing more. In my builds each agent gets its own scoped API key and every tool call is logged. Sending emails, payments and deletions wait for a human click. This costs almost nothing at the prototype stage and is painful to add after something leaks. I cover this in my FAQ on data security.
Why it matters: Before you add the next agent, write down every system it can reach and cut that list to the minimum the task needs.
Source: TechCrunch