Agita Labs was founded on the question of how to build secure computing systems that could adapt to the fast-changing landscape of security threats. In 2017, University of Michigan Professors Todd Austin and Valeria Bertacco conducted a DARPA-sponsored project to build a novel secure CPU that could withstand a wide range of attacks. Their CPU –– Morpheus RISC-V –– would depart from current models focused on finding and fixing software vulnerabilities. Instead, Morpheus supported always-encrypted computation on the program values needed by attackers to craft their exploits. The approach was not only efficient and effective; it made a wide range of security attacks especially difficult to launch.
Initial Morpheus experiments in the lab were stunning: the novel CPU could stop virtually all known control-flow attacks (an attack class that allows attackers to take control of the machine). Building on the early success of Morpheus, Austin and Bertacco launched the startup Agita Labs out of the University of Michigan in November 2018, recruiting UM CSE alum Alex Kisil. The trio then started creating a commercial offering of the Morpheus secure CPU.
The startup was funded by a generous DARPA SSITH commercialization contract. Efforts now focused on the development of Morpheus technology for the Microsoft Azure cloud and also Intel’s high-performance Xeon-FPGA platforms. Concurrently, Agita Labs partnered with Intel on hardware security deployment platforms.
Feedback from potential investors and customers at this stage was uniformly positive: the team not only won the Ann Arbor SPARK “Best of” Award in February 2020; Morpheus also succeeded in being the only system target in the DARPA FETT red-teaming challenge that was not successfully hacked by more than 500 cybersecurity researchers. It was becoming clear that always-encrypted computation was a game changing technology! The team began to consider how Morpheus technology might do more than mitigate software hacking.
If researchers could build a system with generalized always-encrypted computation, it would become possible to create new avenues to share sensitive data! The dream of creating technologies for better privacy moved Austin and Bertacco to pivot in early summer 2020 from building a secure CPU to building and patenting the sequestered encryption enclave –– which is the secure hardware component at the root of all of Agita Labs’ privacy technologies.
For the remainder of 2020, the team hunkered down in their socially-distanced bunkers, building the sequestered encryption hardware enclave, and its software and testing infrastructures. This big build was made possible by Austin’s experience in hardware development and Bertacco’s talent for hardware testing and verification. Simultaneously, the team pursued seed funding, which would allow them to grow their tech and business teams, and advance to a first product offering. By December 2020, Agita Labs successfully closed a $3M seed round that included support from Intel Capital and Walden International.
The next step was to build Agita Labs’ first privacy technology product, which was designed to be deployed into FPGAs on Azure and AWS, plus Intel’s Xeon-Altera CPUs for on-premises deployments. New members of the team were added to start the business team: Sara McLean became the Head of Business Development and Amy Wirth joined as Chief Financial Officer. By October 2021, the tech team successfully ran the first program –– an integer-based bubblesort –– on the sequestered encryption enclave on the Microsoft Azure cloud platform. This was soon followed by countless hours of randomized fuzz testing.
In May 2022, a closed beta release of the TrustForge development platform became available. The platform was deployed into the Microsoft Azure cloud, and it included the TrustForge hardware enclave for Azure, the TrustForge software development kit (SDK) for C++, Python and JavaScript, and the VIP-Bench example benchmarks. Fast forward to now: the team at Agita Labs continues to develop and expand its sequestered encryption technology with the goal of building more secure data sharing technologies for our changing world.
