RSAC 2025: Finding the Signal

RSA Conference 2025 is in the books. It was, as always, a blur — packed halls, packed schedules, product launches, and AI-branded everything. Notably, one vendor even brought live goats to their booth, a spectacle that underscored the lengths companies will go to capture attention.​

But beneath the spectacle, real signals broke through. For cybersecurity founders, especially those building in a capital-constrained but innovation-hungry market, this year’s event clarified what matters — and what doesn’t.

Here’s what we saw, and what we’re taking away.

Capital Flows — But the Bar Is Higher

In the weeks leading up to RSAC 2025, cybersecurity startups announced over $1.7 billion in funding across more than 30 deals — a number that might suggest we’re back in boom times.

But zoom in, and the picture sharpens: the lion’s share of that capital went to just a few later-stage players. ReliaQuest ($500M) and Chainguard ($356M) alone accounted for more than half of the total. Most of the remaining dollars went to companies well past early traction, often with six-figure ACVs and major channel momentum.

The signal for early-stage founders? Capital is flowing — but it’s flowing with intent.

There is real money available at Seed and Series A. But it's going to teams who are solving real, present-tense problems — with clarity, urgency, and precision. Your story has to be sharp. Your architecture has to fit. And your execution needs to get you to customers, not just coverage.

AI Is Becoming the Architecture

Last year, AI was the feature. This year, it’s the foundation.​

The acquisition of Protect AI by Palo Alto Networks marked a turning point. It’s now understood that AI pipelines — models, datasets, APIs, and orchestration layers — represent a new and growing attack surface. Enterprises are treating these systems as first-class assets that must be monitored, governed, and protected.​

At the same time, frameworks like MCP (Model Context Protocol) gained traction as a way to connect LLMs with external tools and data — but not without controversy. Security researchers have flagged potential risks like prompt injection and context poisoning if MCP isn’t tightly controlled. Meta’s new LlamaFirewall and Llama Prompt Guard 2 both reflect a broader shift: building policy enforcement and runtime validation directly into the LLM stack. It’s early, but the direction is clear — security for AI is becoming protocol-level infrastructure, not just an afterthought.

Startups working in this space aren’t just building “AI security tools” — they’re constructing the next layer of enterprise infrastructure: observability, enforcement, validation, and policy for LLM-based systems. And that’s where durable value will be created.​

If you’re building for the AI stack, you’re not too late — but the bar is rising quickly.

AI in the SOC: From Hype to Usefulness

Agentic AI, co-pilots, autonomous analysts — nearly every RSAC booth promised some version of AI-powered SOC transformation.​

Some of it is still noise. But some of it is starting to work.

Big vendors made moves:​

• Google integrated Gemini into Chronicle and Mandiant to generate summaries and investigative support.​

• SentinelOne’s Purple AI took a more Agentic approach — executing chained tasks across systems with reasoning.​

• CrowdStrike unveiled Charlotte AI Agentic Response and Agentic Workflows, extending its AI-native SOC capabilities.

• Cisco launched a Foundation AI Security Model that interprets threats, scores severity, and recommends actions.​

Meanwhile, 10+ other startups are layering AI agents into security workflows: pulling context from EDRs, SIEMs, and identity platforms, then triggering actions within policy constraints.​

But the shift underway isn’t just technical — it’s conceptual. Buyers no longer want more alerts. They want outcomes. Block the attack. Close the case. Fix the exposure.​

The promise of Agentic AI isn’t just autonomy — it’s acceleration. And the winners will be those who build trust with analysts and SOC operators, not just sizzle in demos.​

The Human Layer Gets Operationalized

One of the most quietly powerful themes at RSAC this year: security teams are getting serious about managing human behavior — not just blaming it.​

Human Risk Management (HRM) is emerging as a real category.​

We led the seed round in Amplifier Security, which is turning user engagement into an orchestrated security surface. Think automated patch loops, MFA rollouts, config nudges — delivered as interaction, not interruption.​

BlackCloak, another TechOperators portfolio company, is scaling Digital Executive Protection — safeguarding C-suite and board-level targets who often sit outside traditional perimeter defenses. Just last week, they launched an identity verification feature to help detect deepfake-powered impersonation attacks, reinforcing the need for trust at the human layer.

This category is about more than awareness. It’s about operationalizing user behavior — reducing risk through design, automation, and context. And it’s overdue.​

Identity Gets Its Posture Layer

IAM is finally evolving from access to security.​

Following last year’s attention to ITDR, this year’s RSAC spotlighted Identity Security Posture Management (ISPM). Vendors like Okta, Saviynt, and RSA introduced new capabilities to continuously monitor entitlements, flag misconfigurations, and surface exposure across environments — all without waiting for the next audit cycle.​

Identity is now treated as a dynamic attack surface — not just a directory.

CrowdStrike also expanded its identity protection offerings with Falcon Privileged Access, a module designed to secure the entire identity attack lifecycle across hybrid environments. This solution eliminates standing privileges by granting Just-in-Time (JIT) access based on real-time risk assessments to dynamically grant, block, or revoke access.

For founders, this opens a new frontier: real-time governance, policy control, and hardening of identity infrastructure as a continuous service.​

Final Word: Clarity Over Hype

RSAC 2025 confirmed what many of us suspected: it’s a builder’s market again.​

There’s early-stage capital, but for founders solving real problems — with sharp narratives and fast execution. Buyers are increasingly skeptical of "AI-powered" slogans, and deeply focused on systems that produce outcomes: block, fix, close, or influence behavior. Looking, sounding, or demoing like every other “AI X” isn’t enough anymore.​

The signal? Clarity cuts through. Outcomes matter more than optics. Fit beats flash every time.

And yes — if you’re bringing goats to your booth, make sure the signal cuts through the spectacle.