CCA Shows Why the Autonomy Interface Matters

The Air Force just demonstrated something important with CCA: government ownership of the autonomy interface keeps competition alive.

The public facts are specific. The Air Force selected General Atomics and Anduril for the first Collaborative Combat Aircraft prototypes, now designated YFQ-42A and YFQ-44A. In 2026 reporting, Collins Aerospace’s Sidekick autonomy and Shield AI’s software were tied to semi-autonomous CCA flight work using the Autonomy Government Reference Architecture, or A-GRA. Defense News covered the CCA autonomy software integration across those air vehicle efforts.

That is not just a flight-test milestone. It is an architecture signal. If the autonomy interface is owned by the government and implemented by multiple vendors, the program is not locked into one airframe, one autonomy stack, or one prime contractor’s internal boundary. The government can compare behavior, swap components, and mature autonomy without turning every change into a bespoke integration project.

The alternative is familiar. A program lets each airframe team define the seams around autonomy. The software works in one aircraft because the data model, timing assumptions, safety hooks, and mode logic are local to that implementation. Then the government asks for competition, reuse, or a second platform. The team discovers that the “interface” was really a collection of private assumptions.

An open autonomy interface does not remove the hard work. It moves the hard work to the right place. The standard has to define what state the autonomy receives, what commands it can issue, how authority changes hands, how faults are reported, and how test evidence proves the interface was used correctly. Those details decide whether autonomy can move across compliant platforms.

It also changes how competition can work after the first award. The government can compare autonomy behavior on common scenarios, ask vendors to mature one part of the stack, and avoid treating every air vehicle as a closed vertical product. That gives the program more leverage when mission needs change.

That leverage matters over the life of the program. CCA autonomy will not stop changing after early flight tests. Threats will change, teaming concepts will change, and tactics will change. A government-owned boundary gives the Air Force a path to update autonomy without reopening the entire aircraft integration every time.

This is why A-GRA matters. It gives CCA a government-owned reference point for autonomy integration. It can support vendor competition without asking every vendor to expose its whole design. It can also support air vehicle variation without forcing autonomy vendors to rebuild their software for every aircraft.

At Polyrhythm, we believe this is the right architectural direction for defense autonomy. Our perspective is informed by leadership experience connected to A-GRA’s earliest development. Polyrhythm may be a new company, but not a new voice in this conversation. We see A-GRA as a critical enabler of modularity, interoperability, and long-term competitive resilience in autonomy systems.

The CCA team got this right because the autonomy interface is not a minor software seam. It is the point where air vehicle behavior, mission authority, safety logic, test evidence, and vendor competition all meet. More programs should follow that example before their autonomy stack becomes inseparable from one platform.