A New Tempo for Defense Software

After spending 25 years in defense acquisition, last month marked one year since I joined Shield AI. It feels like an appropriate time to reflect on our work on the U.S. Air Force’s Collaborative Combat Aircraft (CCA) program, what that work reveals about the changing role of software in defense, and how government and industry can meet SECWAR’s call to accelerate delivery of capability to the warfighter.

Before I arrived, the Air Force had selected Shield AI to integrate Hivemind mission autonomy software onto its CCA. These uncrewed aircraft are designed to operate alongside crewed fighters under the direction of human operators. By January 2026, Hivemind was flight ready. In March, it piloted the YFQ-44A across three sorties and more than four hours of flight time. The campaign exceeded the government’s initial objectives and matched the performance predicted by Hivemind’s digital models.

Since then, the team has built a backlog of flight-ready capabilities while remaining compliant with the Autonomy Government Reference Architecture (A-GRA). That work reflects a core conviction: we must fully exploit the government’s deliberate separation of safety-critical aircraft functions from the mission autonomy software, an architectural choice intended to enable faster capability development, integration, and deployment.

This is not only a CCA story. Software is moving faster across defense. The processes used to integrate, test, approve, and field it must evolve to keep pace.

Software Is No Longer the Long Pole

Throughout my acquisition career, hardware programs usually waited on software, and software almost always carried the biggest risk. On CCA, that order has changed. Our development pipeline runs continuously, delivery has compressed from months to days, and the team holds a flight-ready backlog. The next step is one we must take together: establish a trusted, repeatable path for integrating, testing, approving, and releasing mission autonomy to fleets.

Moving fast does not mean cutting corners. In fact, it improves quality by giving acquisition teams more opportunities to find and solve hard problems early. Operator testing, program reviews, integration events, safety processes, and airworthiness reviews all remain in place. Testing in simulation and flight builds technical confidence and gives operators and acquirers more evidence to evaluate performance. The goal is disciplined speed: build early, test continuously, and learn from each result.

Hivemind gains momentum with each program implementation because it was built to integrate horizontally. It has been integrated on more than 40 uncrewed systems across air, land, sea, and space, with every implementation starting from a common foundation. Tools, interfaces, safety evidence, and integration patterns carry forward, lowering risk on the next one.

This is the promise behind software sold separately. Manufacturers build the platform. Mission autonomy providers continuously improve the intelligence that flies it and the evidence needed to field it. Realizing that promise requires government and industry to align policy, infrastructure, and capacity around a repeatable release path.

Trust Is Earned in the Technical Trenches

The March 2026 campaign validated performance first shown four months earlier in high-fidelity laboratories with Air Force users. Hivemind controlled the YFQ-44A, demonstrated basic navigation, stayed inside its assigned boundaries, and worked reliably through a human-machine interface across multiple mission phases.

A recent Agile Combat Employment exercise built more trust in Hivemind. For the first time, an Air Force pilot used mission autonomy to direct two beyond-visual-range combat engagements while thunderstorms surrounded the range. The historic mission reinforced our digital models and led to fist bumps across Shield AI and government teammates.

Flying the software is what turns predicted performance into valuable evidence. Each sortie gives operators greater confidence, validates the digital models used before flight, and expands a body of evidence that lowers risk for the next integration event.

That accumulating evidence is how trust is earned—one software release at a time. Every laboratory run, safety review, and flight must build confidence not only in the software, but also in the process used to deliver it. That confidence makes fly-update-fly possible at the edge: update, verify, release, and fly again within hours—without starting from zero each time.

What Must Change to Turn Ready Software into Fielded Capability

The Air Force’s Software, Autonomy, and Teaming (SWAT) branch at Wright Patterson Air Force Base, Ohio, has set a standard for disciplined urgency and extreme teaming with key stakeholders. That same spirit should apply across all defense programs. The platforms may differ, but the barriers to fielding mission autonomy are often the same. Five changes could help.

1. Write mission autonomy into test strategies from the start
   Avoid testing the platform first and then mission autonomy later. That validates the hardware but delays learning about the complete system. If autonomy is part of the operational workflow, it should be included in the first relevant test event.
2. Stand up independent laboratory capacity
   Mission autonomy providers need access to current hardware, interfaces, and digital models. A dedicated hardware laboratory at the mission autonomy supplier’s location would allow mission software to be enhanced, tested repeatedly, fielded immediately (pending approval) without waiting behind the larger platform development schedule.
3. Add test capacity dedicated to mission autonomy
   Software cannot improve through real-world evidence if there are too few opportunities to test it. Programs need access to more platforms, better range space, and key support personnel when the software is ready.
4. Modernize infrastructure for collaborative autonomy
   Testing several autonomous systems together requires more than infrastructure designed for individual platforms. Modern range safety, communications, and data collection systems (e.g., SkyVision) can provide collaborative testing more test capacity.
5. Use digital evidence to accelerate approvals
   The goal is not less scrutiny. It is using evidence from modeling and simulation, laboratory testing, and real-world performance to focus airworthiness and safety reviews on what actually changed. Together, we can establish a trusted release process with defined bounds for change—so every update does not restart the approval cycle.

These recommendations will require buy-in from program offices, test and airworthiness organizations, platform manufacturers, and autonomy providers.  Given SECWAR’s demand to accelerate capability and capacity, I believe the time to move out on them is now.

The Next Chapter

In less than a year, Hivemind went from laboratory integration to flight aboard a CCA, and the team keeps delivering capability ahead of the next test opportunity. The same pace applies beyond one aircraft or one military service. Mission autonomy is moving onto uncrewed systems across air, land, sea, and space.

Defense leaders want faster delivery and more capacity. Shield AI is working with government and industry partners on both. Let’s band together and establish a shared operating tempo across software delivery, test infrastructure, aircraft integration, and approvals—so mission autonomy can be tested, updated, and fielded at the speed the mission requires.

About the author:

Tim Fuhrman is the Senior Director of Strategic Systems at Shield AI, where he is responsible for the execution and growth of our Combat Air, Strike, and Penetrating ISR domains. Before joining Shield AI, Tim spent over 25 years in the U.S. Air Force leading large acquisitions and aircraft maintenance units. He retired in 2025, serving his last 3 years as Senior Materiel Leader for the Adaptive Weapons Division at Eglin Air Force Base, overseeing an $18B portfolio of strike weapons programs while pioneering the affordable mass movement known as M-Series weapons. Tim holds a B.B.A. from the University of Georgia’s Terry College of Business and an MBA in Finance from Auburn University at Montgomery.