At RSA Conference last week, Cisco dropped a set of announcements that the trade press dutifully covered as a product launch. That framing misses the point entirely.
What Cisco actually did in San Francisco was declare where the security partner market is going over the next three years — and start building the on-ramp for partners who want to get there. For MSSPs paying attention, it’s a significant opening. For those waiting to see how the market shakes out, the window is shorter than they think.
The stat that should scare every security partner
Cisco released survey data alongside its RSAC announcements: 85% of enterprises are experimenting with AI agents, but only 5% have moved them to production. That number tells you exactly where the money is sitting right now. Nineteen out of twenty enterprises want to deploy agents and can’t. The bottleneck is security — identity, access control, governance, trust. These aren’t unsolvable problems. They’re exactly the problems MSSPs were built to solve for customers who won’t figure them out on their own.
Cisco is betting that unlocking production AI deployment is the next major security services market. The announcements at RSAC are infrastructure for that bet.
What actually shipped
The centerpiece is Zero Trust Access extended to AI agents — which sounds theoretical until you map it to real customer problems. Most enterprises today have no idea which agents are running in their environment, who authorized them, or what data they can touch. Cisco’s answer is Agent Identity Management through Duo IAM: register agents, map them to accountable human owners, assign fine-grained permissions scoped to specific tasks and time windows. All agent traffic routes through an MCP gateway so nothing goes dark.
That’s not a Cisco product feature. It’s a managed service waiting to be built.
Cisco also introduced DefenseClaw, an open-source secure agent framework that automates security inventory for AI workloads, with planned integration into NVIDIA OpenShell. And AI Defense: Explorer Edition gives developers self-serve tools to test agent resilience against attacks before deployment — a product that plugs directly into a pre-deployment security review service.
On the SOC side, Splunk’s new automation capabilities target response workflows at machine speed. The Automation Builder Agent and Triage Agent are expected to ship in June. Detection Builder and Guided Response Agent are in prerelease testing.
The real channel play
The existing MSSP model built around endpoint monitoring and ticket-based incident response is already under margin pressure. What Cisco outlined at RSAC is essentially the successor model: AI agent governance as a managed service.
Think about what that looks like in practice. A customer wants to deploy ten AI agents across their organization — one for HR workflows, one for IT triage, one for customer-facing support. Before a single agent touches production, someone needs to register identities, establish ownership chains, configure access scopes, run adversarial testing, and stand up monitoring. After deployment, someone needs to watch for behavioral drift, manage policy updates, and respond when something goes sideways.
That’s not a one-time professional services engagement. It’s a recurring managed service with natural expansion as customers deploy more agents.
Jeremy Nelson, CISO North America at Insight — one of the larger solution providers operating in the security space — flagged the same read in Cisco’s RSAC press materials: “Organizations are eager to embrace AI, but they need to do so without creating security coverage gaps.”
That’s a customer who needs an MSSP. Not a product vendor.
The window is narrower than it looks
This story follows a familiar pattern in the channel. A technology category opens up — cloud security, zero trust, MDR — and there’s a 12-to-24-month window where early-mover partners capture the best customers, build the best playbooks, and become the established choice. After that window closes, the market consolidates and the late entrants compete on price.
The agentic AI governance window is open right now. Cisco has the broadest platform for it (Identity + SSE + SIEM through Splunk). The market has a specific, documentable problem (85% testing, 5% production). And enterprises are conditioned to trust their existing security MSSP to solve it.
Security partners who want a position in this market need to move in Q2. Not because the technology is mature — it isn’t yet. But because the customers who deploy first will look to the partner who helped them first when they need to scale.
The 5% in production today will be 30% by end of 2027. The question isn’t whether enterprise AI agent deployment takes off. It’s which MSSPs are ready when it does.
Those Automation Builder and Triage Agents ship in June. Start building your practice around them now.
See also: Cisco’s Splunk acquisition and what it means for security partners | Why your security practice needs an AI governance layer