The Mandos platform tracks 44 companies with an AI red teaming product, and together they've raised about $1B. That's the group Dreadnode comes from, the folks who attack a model or an agent on purpose so its owner learns what breaks before a customer does. CrowdStrike now owns one of them, and it paired it with Fiddler, which records and explains what agents do once they're live.
My guess on why the money's going here: an agent doesn't do the same thing twice, so you can't sign it off like a firewall rule. The honest answer to "is it safe" ends up being test results and a log. Sentra gives you that for data, Brava for the SOC.
And Palo Alto Networks is reportedly talking to Clutch Security, which tracks the keys and identities agents use, about a deal around $500M (TechCrunch, citing The Information). So I wouldn't be surprised to see this again soon.
For founders, my gut says the product that wins the next renewal leaves evidence behind, a test report, a trace, an access map someone can hand to a board. "We stop it" is a tougher sell in a budget meeting because nobody can see what got stopped. For investors, both exits look good (Fiddler raised about $100M and went for a reported $400M, Dreadnode had a $14M Series A and went for a reported $95M), but the list of buyers who can pay that is still pretty short.
And if you run security, just make sure you can export those test results and traces into your own systems and keep them after the contract ends. The vendor holding them might also be the one on your endpoints.
Could be wrong here. Two deals from one buyer in one week might just be CrowdStrike's strategy, not a trend, and both prices are reported, not disclosed. Still, I'd build for the showing.
One question
If you build or back an AI security company: what are your customers actually paying for first? One click is enough.
Or just hit reply and tell me why, one line is plenty. I read every one, and the best answers go in next week's issue, anonymously.
- Dropzone AI rewrote its pricing page between 29 October and 5 November. "Contact us for enterprise pricing" is gone, and the new headline reads "A SOC works as a whole. We price it as one.", with the line "Enterprise is sized on employees; federal and MSSP are priced to how you run."
- Didit changed its pricing between 4 and 7 November, from "Three tiers, one price list" with "500 free verifications every month, forever" to "Four plans, one account" with "$10 of free credit every month". It also added a Growth plan at $99 a month and a Scale plan at $299 a month.
- Darktrace swapped its homepage headline "BEHAVIORAL SECURITY IN THE AGE OF AI" for "You can secure what AI can do" between 4 and 8 November, and a new line promises "behavioral security that understands agent activity".
- Native Security changed its homepage line between 3 and 6 November from "Native turns your built-in cloud security controls into operationalized secure architecture" to "Native operationalizes your security architecture, turning intent into preventive enforcement across your cloud and AI infrastructure."
- SentiLink added a third ending to its rotating headline between 4 and 7 November, so "Your partner in solving identity theft, synthetic fraud" now also reads "emerging threats".
- Aliro Quantum changed "the foundation for Entanglement as a Service" to "the foundation for scalable quantum networking" between 4 and 7 November, and "Use entanglement today" is now "Build your entanglement-based network today".
Pacesetters watch
What the companies in my quarterly Pacesetters ranking did this week:
- Sardine (#1 of 40 in Fraud & Payment Security) launched Know Your Human on 5 November, a protocol that lets a merchant see the real shopper behind an AI agent at checkout. Agents again, just from the fraud side.
- Aikido Security (#3 of 97 in Application Security) launched Deep Review on 2 November, where swarms of AI agents review every pull request and post fixes straight to it.
- Abnormal AI (#4 of 24 in Email & Messaging Security) now sells AI Governance as a standalone product, no email security needed, and named Don Jeter as CMO on 4 November. Pretty clear where they want to grow.
- Horizon3.ai (#1 of 39 in Offensive Security) is hiring 50+ people across APJ, and Terra Security (#7 in the same category) hired Tal Kapulnik from Eon as CMO. Offensive security is staffing up.
- Tailscale (#2 of 72 in Network Security) welcomed 19 new people in one post on 2 November.
Thanks for reading.
If someone you know builds or backs security companies, forward this to them, it's how the Brief grows.
Talk to you next Monday,
Nikoloz
P.S. I post market reads most weekdays on LinkedIn, come say hi there.
