Margin of Safety #12 – 2025 RSA Takeaways

Kathryn and I both spent the better part of last week at RSA. We will cover our three biggest takeaways: 1) Chainguard 2) AI fluff and 3) Wiz-ification.

1. Chainguard’s RSA Moment: Momentum, Questions, and the Shape of What’s Next

If you walked around RSA this year, it was hard to miss Chainguard. Their presence was sprawled across the Four Seasons, Market Street, and of course, the Moscone floor. It wasn’t just the purple octopus branding — it was a signal. Fresh off a $356M Series D, reportedly raised at a 50–60x+ ARR multiple, the company & investors are betting big that its approach to securing the software supply chain isn’t just nifty — it’s something of a game changer.

There’s no doubt the product is innovative. At its core, Chainguard rethinks vulnerability management to deliver hardened, continuously patched container images. But the real go-to-market unlock has been compliance. For buyers grappling with FedRAMP, CMMC, or other alphabet soup mandates, Chainguard turns security hygiene into a procurement accelerator.

And yet, beneath the RSA shine, a few question marks are starting to show. In conversations with CISOs, there’s a recurring unease: is Chainguard truly eliminating vulnerabilities, or simply ensuring vulnerabilities evade detection by scanners like Wiz?

Meanwhile, competition in the container hardening and supply chain security space is accelerating. Root, RapidFort, and others are aggressively going after the space. As the market expands, questions of differentiation increasingly hinge on deployment friction, ecosystem integration, and proof of efficacy—especially in real-world incidents.

The underlying business model is very interesting. This category is not the traditional asset-lite SaaS model, but instead requires significant upfront investment in the large repository of clean images or packages (plus the staffing to deliver on an SLA for remediating newly discovered issues). This requires strong fixed cost leverage to achieve software-like gross margins, but has the potential to surpass traditional margins if and only if a provider can achieve critical mass adoption. But high multiple bets require not just revenue scale, but narrative clarity (what’s after container images to get to $200M+ of ARR?). Chainguard has the spotlight—now it must prove that its approach is not just novel, but enduring.

2. AI Fluff!

Another theme was AI agents. We lost count of the number of agentic security automation startups we saw, and even classic companies are touting their agentic capabilities. In some cases, the value prop is clear – the rush toward SOC agents is driven by clear business needs for better results at fixed headcount – but in other cases, it’s ambiguous who the intended market is.

We are genuinely curious if any buyers are specifically seeking agentic capabilities, or if the real buying criteria continues to be product capabilities and total cost of ownership, with agents being used as a lazy shorthand to imply low TCOs. To the extent buying criteria aren’t changing, there seems to be a marketing opportunity for the players who can genuinely point to realized TCO and capability gains, versus leading with an increasingly noisy AI storyline. In a world where agent has come to mean everything and nothing, we think the real differentiation will come from customer impact rather than technical architecture. As one CISO put it to us, “Don’t tell me you have agents. Show me how I can fire three vendors.”

3. Wiz-ification of security startups

Beyond the hype and headlines, one quieter but more encouraging trend stood out: more startups are internalizing a key lesson from Wiz’s rise — that great UX directly translates to better usability, lower cost of ownership, and happy customers.

It’s not just about dashboards that look nice (though good looks and clean presentation always help users find key details). It’s about building products that balance useful high-level summaries with the key insights a practioner needs, without requiring them to work for it. Several startups we saw — from posture management to third-party risk — are embracing progressive feature discovery, surfacing complexity only when needed and guiding users with clear actions rather than another compliance checklist. It’s the difference between being a process tracking spreadsheet and being a useful system of record.

Critically, the teams getting this right aren’t building another security co-pilot — they’re building verticalized experiences that natively fit into practitioners’ days. We saw products that mapped security findings to real developer workflows, that exposed just the right alert metadata for triage, and that skipped the AI monologue in favor of “here’s what changed, here’s what we recommend.”

It’s early, but promising. Not every startup needs to / will become a Wiz. But more seem to be absorbing its deeper lesson: when you combine clean, high-signal data with a user experience that earns trust and reduces work, you don’t need to shout about your AI agents. Users just come back — and bring others with them.

If we missed you at RSA, please feel free to reach out!

Kathryn Shih – kshih@forgepointcap.com

Jimmy Park – jpark@forgepointcap.com

Stay tuned for more insights on securing agentic systems. If you’re a startup building in this space, we would love to meet you. You can reach us directly at: kshih@forgepointcap.com and jpark@forgepointcap.com.