2026.07.30
Evaluating AI for Security and Intelligence Teams.
A candid conversation on when to trust AI for intelligence work, and when not to.
Most intelligence teams are already using AI. The question is now whether you know when to trust it.
We ran 18 real intelligence questions through six leading AI models and scored every answer: situation reports, pre-deployment briefs, scenario forecasts, with false premises and fabrication traps built in. What we found is the uncomfortable headline: the models are good enough often enough that most teams have stopped asking when they aren't. That gap is where the risk lives.
This session walks you through what we found, then turns the question on the room. You'll leave knowing which tasks to trust a general model with, which to keep away from it, how to prompt for output you could put in front of leadership, and how to start building your own workflows — whether or not you write a line of code.
Recap
Key takeaways
- The difference between a general model that's good and one that's trustworthy is architectural. Know when prompting can and cannot bridge that gap.
- Think about the enterprise tools you're being provided. Be intentional about what you are willing to use them for. Understand their left and right limitations.
- Before you evaluate any AI intelligence tool, ask what's in it, what's excluded, and who verified it. Today, the corpus is the product.
- Consistency matters more than peak performance. A model that applies analytical standards selectively is not reliable — even when individual outputs look right.
- Governance that lives in a policy document but not in the product is a risk that can be managed — but it should be recognized as a risk.
"One thing I want to start with is something I hear constantly in conversations across the industry right now. It usually sounds something like this: 'We got access to an AI tool — is it good enough for what we need?'"
For the full conversation, watch the recording above.
Decision advantage at the speed of escalation
See proactive intelligence in action On your own risk picture.
Seerist turns verified events and expert analysis into decision-ready intelligence for your team, on demand and in real time.