Securing the AI Agents
Sai Praveen@saipraveen9999
−11.5%
since Oct 13 · after money added
The idea
I believe AI agents logging into systems with real credentials and little oversight will push security budgets higher over the next 12 to 18 months, concentrated in identity, traffic control, and data protection. To express this I lean toward smaller, faster-growing names across those layers, with only two larger holdings anchoring the basket, rather than betting on a single company. About a tenth sits in cash to add if prices drop, since a broader software selloff could still drag these stocks down even if the underlying idea plays out.
What it owns
- These companies see and control what agents actually do on devices and across the network, where a rogue agent gets caught. · Cloudflare, Inc.It sits in front of web traffic and can distinguish agents from people, deciding what gets through while providing the developer platform where agents run., CrowdStrike Holdings, Inc.It watches what software does on company machines and servers in real time, detecting agent activity and unwanted behavior as it happens., Zscaler, Inc.It routes and inspects every connection between users and apps without trusting anyone by default, a model that applies to agent traffic as well., F5, Inc.It sits in front of company apps and inspects the traffic agents send when calling those apps, controlling and securing agent access points., Akamai Technologies, Inc.It blocks bad bots and protects the connection points agents use to reach company systems, providing security at smaller scale., Fortinet, Inc.It sells firewalls and network security that companies need as agent deployment grows, steadying security spending regardless of AI spending changes.33%
- Every agent becomes a new user that needs a login and limited permissions, so identity companies sell more as agents spread. · Okta, Inc.It provides the login and identity layer that controls what agents can access within company apps and is pushing phishing-resistant authentication for the AI era., Palo Alto Networks, Inc.It protects the privileged logins that agents use to reach sensitive systems, combining network security with the access controls from its CyberArk acquisition., SailPoint, Inc.It decides what each account can touch and removes access nobody needs, keeping agents from reaching data beyond their intended scope.24%
- When AI agents access restricted data or cause system failures, these companies limit damage and restore lost information. · Rubrik, Inc.It backs up company data and rolls it back if agents or attacks cause damage, and now includes tools built on Claude to undo agent-caused harm., Varonis Systems, Inc.It maps file access permissions and detects when accounts pull data abnormally, which becomes more critical as AI agents gain broad read access to company systems., Commvault Systems, Inc.It backs up company data and helps recover after attacks, providing the cleanup capability needed if agents or attackers cause damage.20%
- These companies find the holes agents open up and spot attacks early, which matters more as agents widen what attackers can reach. · Tenable Holdings, Inc.It finds weak spots across company systems that agents and AI tools might exploit, helping companies stay ahead of new vulnerabilities., SentinelOne, Inc.It uses AI to spot attacks on company machines in real time, serving as a smaller alternative for detecting agent-based threats., Qualys, Inc.It scans systems for known security holes, work that increases as agents add more things to check and potential vulnerabilities to assess., Rapid7, Inc.It sells detection and vulnerability tools that help companies find weak spots agents might exploit, with potential to grow as agent security concerns rise.13%
- Cash10%