Calypso’s Red-Teaming and Agentic Threat Tools Boost F5’s Application Security Edge
F5 plans to purchase an AI security startup led by a longtime Qualtrics executive to secure modern application environments in the context of emerging artificial intelligence workloads.
See Also: Post-Quantum Cryptography – A Fundamental Pillar in the Future of Cybersecurity [ES]
The Seattle-based application and API security vendor said Dublin-based CalypsoAI’s AI guardrails, red-teaming capabilities and platform flexibility for deployment across SaaS, cloud and on-prem settings stood out among the crowd. CalypsoAI’s technology enables deep conversational analysis, identifying prompt-level security violations and addressing emerging challenges including shadow AI and data leakage.
“A big part of what drew us to Calypso was that they were very, very much aligned with the way that we at F5 think about our customers and where they’re deploying their workloads,” said Senior Vice President of Product Management Shawn Wormke. “The way that their approach married with our views of the application delivery security platform made it a very natural for us.”
CalypsoAI, established in 2018, employs 50 people and has raised $43.2 million in outside funding, having most recently received $5 million from Crosspoint Capital Partners in April 2025 in conjunction with being a top-two finalist at the RSAC Innovation Sandbox. The company has been led since August 2024 by Donnchadh Casey, who spent nearly nine years at Qualtrics, ending in the role of chief customer officer.
F5 had already launched its AI Gateway offering, which saw early success in customer environments, Wormke said. This created an imperative to expand into more robust AI security capabilities, specifically at the point of inference, where security risks are most prevalent in AI applications, he said (see: F5 Buys Startup Fletch to Automate Security With Agentic AI).
“The efficacy of the solution that they have, the performance of the solution that they have and the approach that they were taking to their AI guardrails capabilities, along with their red-teaming capabilities, and how they’re able to use that to enhance and improve their security solution overall was appealing,” Wormke said. “They were very much aligned with our hybrid and multi-cloud approach.”
How F5, CalypsoAI Will Come Together
Wormke said Calypso’s strength lies in its agentic warfare tactics and red-teaming methodology, with their systems constantly attacking their own models to find vulnerabilities and feeding those insights back into their AI firewall to improve defenses in near real-time. This continuous, closed-loop approach to threat discovery gives Calypso a technological edge and aligns with F5’s emphasis on app security.
“It’s the approach that they take to their agentic warfare and red-teaming, and the way that that feeds back into their system,” Wormke said. “We think it’s the right base to build the world’s leading AI security solution.”
Initially, F5 will enable traffic routing to the Calypso platform, allowing existing customers to tap into its capabilities without big changes to infrastructure. This will evolve into a deep native integration, where Calypso’s technology becomes part of the actual F5 runtime platform rather than just something bolted on. The objective is to make AI security invisible to the customer from a deployment standpoint, he said.
“That becomes just a more and more native part of the platform as the technology evolves,” Wormke said. “So, it becomes a native experience for our customers, so it’s seamless for them and very easy for them to turn on and acquire that technology.”
Calypso’s agentic red-teaming identifies over 10,000 new attack types per month, and this intelligence is directly fed into their AI firewall. That makes Calypso unique in its ability to defend against new and evolving threats. Calypso offers mechanisms to block or scrub sensitive data in real time as it flows through AI models, including preventing models from responding with confidential or regulatd data.
“We’ll also be able to integrate that into the Calypso solution, giving us a very effective data leakage prevention solution for AI prompts,” Wormke said.