The Year of Agentic AI at RSAC 2025—Cutting Through the Noise

RSA Conference 2025 by the numbers

RSA Conference (RSAC) 2025 drew more than 43,500 security professionals, 730 speakers, and 650 exhibitors at San Francisco’s Moscone Center—a record crowd that heard one dominant theme throughout the four day event: Agentic AI—autonomous systems that observe, reason, and act on live data—is the next frontier of cyber defense. In fact, Agentic AI was the most-tracked term “by far,” eclipsing every other buzzword at the event. 

Agentic AI is everywhere, but execution matters

From the opening keynotes to the hallway chatter, the topic of agentic AI dominated every conversation. It became resoundingly clear that, now, any vendor claiming agentic AI capabilities must be able to back it up and prove their worth.  

In other words, simply layering a chatbot on a dashboard no longer impresses. Now the litmus test is whether autonomous systems can achieve actual security outcomes. Can they surface previously unseen exploits? Can they slash triage cycles in the Security Operations Center (SOC)? Can they fuse cyber and physical signals fast enough to prevent real business disruption? Whatever the use case, tangible outcomes must follow.

Dataminr’s approach: Purpose-built agents for actionable intelligence

We answered the call unveiling Dataminr Intel Agents, the first milestone on our Agentic AI roadmap. Dataminr Intel Agents continuously synthesize and correlate cyber signals from our knowledge graph to enrich Dataminr Pulse for Cyber Risk alerts with deep, real-time, and mission-critical context (i.e., insights that can take SOC analysts hours to piece together manually on their own). Select clients in private beta are already leveraging Dataminr’s AI-native platform and Dataminr Intel Agents to:

1. Unearth emerging threats and zero-days before they strike. Multimodal fusion AI fuses audio, video, image and text data to identify early and anomalous threat activity in real time—often hours or days or weeks—before legacy threat intelligence tools can. By surfacing only relevant alerts from the vast noise of publicly available data, security teams gain a critical window to assess risk, mobilize resources, and mitigate impact before threats escalate.
2. Triage and respond to attacks at machine speed. Autonomous threat correlation and enrichment provide security teams with deep, instant insights into active exploits, ransomware events, threat actor tactics, techniques, and procedures (TTPs), third-party supply chain attacks, and more. 
3. Maintain real-time situational awareness with live updates. Continuous multimodal collection and agentic summarization keep analysts and executives aligned on threat evolution—automatically updating alerts with new context as cyber events unfold.
4. Monitor the third-party attack surface continuously at scale. Equipped with visibility into global supply chain disruptions, partner vulnerabilities, and vendor-related incidents, security teams can continuously monitor and mitigate third-party risks in real time at scale across their entire third-party ecosystem and prevent supply chain threats from impacting core operations.
5. Address converged cyber-physical risk events. Dataminr stitches together cyber telemetry with geopolitical and physical security feeds for converged intelligence. This enables cyber and physical task forces to coordinate mitigations, protect personnel, and safeguard uptime in unison—addressing the response in concert—whether it’s a forest fire, protest, IT outage, or ransomware attack.

Cut through the noise and supercharge your SOC

Even though Agentic AI skyrocketed as the cybersecurity buzzword of the conference this year, it doesn’t mean the term is a moonshot and all hype. To get beyond the showfloor hype and actually determine Agentic AI fact from fiction:

• Forget the tech, focus on outcomes instead. Insist on quantifiable evidence and real-world customer examples that demonstrate how Agentic AI reduces Mean-Time-To-Remediate (MTTR), dwell time, and false-positive rates. Ask for independent validations, side-by-side baselines, and customer references that translate technical gains into business resilience—downtime avoided, fines mitigated, and analyst hours reclaimed.
  
• Scrutinize the data foundation. Agentic reasoning is only as strong as its inputs. Press vendors on the breadth, freshness, and provenance of their data pipelines: Are they ingesting multimodal signals—text, imagery, network flows, and dark-web intelligence—in real time? Without a robust data backbone, even the smartest agent is flying blind.
  
• Measure the business impact early and often. Pilot the tools yourself and establish a crystal-clear success rubric. Track baseline metrics on day zero, then instrument automated scorecards that report weekly deltas in response speed, incident volume, and analyst workload. Force the platform to earn its budget every sprint; continuous measurement keeps marketing claims honest and accelerates ROI realization.

Looking ahead

Ready to experience agentic AI in your own environment? Schedule a demo of Dataminr Pulse for Cyber Risk today.