By George Kaminsky

Shield AI will put its Hivemind autonomy software into a family of Australian-designed loitering munitions, a deal that points to where military AI is heading: not a human steering each aircraft, but software that plans the mission, coordinates several machines and keeps working when the signal is lost. The San Diego company announced on October 6 that it had signed a memorandum of understanding with Perth-based Innovaero Technologies, at the Land Forces 2026 exhibition in Perth, to integrate Hivemind into Innovaero’s One-Way Loitering, or OWL, munitions.

A memorandum of understanding is an agreement to cooperate, not a purchase. Neither company has announced a contract value, a customer order or a fielding date.

Shield AI describes Hivemind as mission autonomy software, the layer that lets an aircraft operate without a pilot and without relying on GPS or a communications link. In this arrangement the company says it will help Australian forces plan and execute missions faster, coordinate multiple aircraft and keep operating when navigation or communications are degraded. Innovaero says the integration will speed mission planning and increase the lethality of OWL attacks.

Those three claims describe the same shift from different angles. Planning is the human-intensive step today, because someone must decide routes, timing and targets. Coordination turns several single-use aircraft into a group with a shared plan. Operating through jamming matters because degraded signals are an expected condition on a modern battlefield, not an edge case.

Innovaero’s OWL line has three members at different stages. The OWL-B is an electric-powered munition that the company began developing in 2022 with the Australian Defence Force for medium-range strike. The jet-powered OWL-X is under development. The long-range OWL-C remains in the design phase.

Shield AI says Hivemind has already been installed on the OWL-B ahead of upcoming flight tests. The software is being used to simulate coordinated, multi-aircraft strike operations in changing scenarios, with operators directing missions through the Android Team Awareness Kit, a widely used tactical interface. That detail matters for the oversight question: the companies describe a human directing the mission through a tactical application, but they have not published how much discretion the software would have in selecting or engaging targets.

Earlier reporting gave the OWL-B a range approaching 200 kilometers, or about 30 minutes of loiter at 100 kilometers, with anti-armor and fragmentation warheads up to 7 kilograms and a launch weight near 30 kilograms. Those figures come from earlier coverage of the program, and neither company repeated them in the latest announcement. Innovaero has also been testing the OWL-B from the flight deck of a training vessel, a step toward possible shipboard use by the Royal Australian Navy.

Shield AI is leaning on its record of fitting the same software onto different airframes. It says it most recently integrated Hivemind onto the Destinus Hornet loitering munition within two months, with Australian teams supporting successful flight tests in Spain. The pitch to Innovaero follows the same logic: bring proven autonomy software to an airframe whose maker has deep experience with the hardware.

Adam Hogan, the Shield AI vice president who leads Hivemind for Australia, Asia and the Middle East, tied the deal to the idea of sovereign capability, saying Australia gets greater control over how the autonomy is developed, tested and fielded. That framing matters to allied buyers who want munitions built at home but do not want to build their own autonomy stack.

Loitering munitions have moved from niche weapons to standard kit in recent conflicts, and the next competitive step is the software that lets many of them work together without constant supervision. Putting that software on a locally built munition shows how a U.S. autonomy company can extend its reach through partners rather than exports of finished hardware.

It also lands as Washington reorganizes around the same idea. The Pentagon has announced plans for an Autonomous Warfare Command, and allied militaries are watching which software they will end up flying.

The key markers are measurable. First comes the OWL-B flight test with Hivemind aboard, which would show whether the simulated multi-aircraft coordination carries over to real aircraft. Second is any Australian Defence Force commitment, such as a funded integration program or an order. Third is detail about human control: how operators approve targets, what happens when links drop, and which functions stay with a person.

Until those appear, the agreement is a statement of intent from two companies with credible products, and the testing schedule, not the announcement, will show whether the software performs as described on a real munition. It shows the direction of travel, toward coordinated, jam-tolerant munitions, but not yet a fielded capability. For allied buyers, the practical question is whether autonomy can be fielded quickly on a locally built munition without turning into a dependency on a single foreign supplier.

Source: Shield AI and Innovaero announcements as reported by Janes, Defence Connect and The Defense Post, October 6, 2026; Janes, November 2025.