Skip to main content
AI & Machine LearningQuantum Computing

Air Force Seeks AI to Streamline Minuteman III Missile Fleet Data

Air Force personnel surrounded by computer consoles and monitoring screens in a Minuteman III control room.

"approximately 60 disparate and disconnected systems," the Air Force wrote in a July 24 request for information, underscoring a simple technical problem with grave operational stakes: the service cannot centrally govern the many data sources that underpin its aging Minuteman III intercontinental ballistic missile (ICBM) fleet.

What the Air Force asked for in its July 24 notice

The request for information published July 24 seeks an artificial-intelligence agent to create a “single, unified interface” that can pull together data from a range of sources supporting the Minuteman III (MMIII) sustainment lifecycle. The notice says MMIII sustainment currently relies on data scattered across “approximately 60 disparate and disconnected systems,” including everything from local hard drives to shared networks, and that the program has lacked a way to centrally govern those sources.

The notice frames sustainment as a national security imperative: the Air Force must continuously keep 400 missiles ready to launch at a moment’s notice, a responsibility the notice describes as a “no-fail” mission. The RFI also notes that MMIII must be sustained longer than originally planned because the replacement program, Sentinel, faces years of delays.

What the requested AI agent would be expected to do

The Air Force describes several concrete capabilities it wants from an AI tool. It should be able to construct a “single, unified interface” that pulls data from design drawings, maintenance records and other areas and visualizes that information for users. The tool is expected to collate data in “near-real time” and reduce manual compilation by a “measurable amount,” a performance target the Air Force says will be negotiated with the successful contractor.

The notice also acknowledges prior attempts to create a centralized digital ecosystem for the MMIII lifecycle have failed “due to a range of challenges,” and sets a staged approach: the project’s “initial phase” will handle data marked as controlled unclassified information (CUI), with the possibility of later expanding to information classified as secret.

Operational limits and human control

The notice explicitly does not propose giving AI any role in launching or operating the nuclear missile fleet. It says officials maintain launch and operational decisions belong in the hands of humans, signaling that the sought AI agent is intended for data integration, visualization and sustainment support rather than command or control.

Industry context: the Air contract and recent Air Force AI demos

Prior to the RFI, the Pentagon awarded Air — the firm formerly known as Govini — $31 million for work supporting the Air Force’s ICBM fleet. In a July 15 press release, Air said that contract work would “provide an integrated view of the ICBM industrial base and supply chains to identify current and future supply chain gaps and assess risks.” The RFI does not specify how, or whether, Air’s platform would interact with the AI agent the Air Force is seeking.

The notice appears amid broader Defense Department and Air Force moves to experiment with AI. The Defense Department under Secretary Pete Hegseth has been championing new AI tools and searching for hundreds of software engineers to expand its software capabilities. The Air Force on Aug. 4 described a milestone using a heavily modified autonomous F-16 called the X-62 VISTA: an aircraft AI agent analyzed live infrared sensor data and directed the jet to “intercept an airborne target in real time.” A separate demo incorporated a “non-prime, third-party autonomous agent” from an undisclosed vendor to test safety rules.

What this means for technologists, procurement leaders, and the public

  • Technologists and security teams: They will face the concrete technical task of integrating data across roughly 60 disconnected systems, initially handling CUI and potentially moving to secret data, and delivering near-real-time collation and visualization while avoiding past pitfalls that derailed earlier centralization efforts.
  • Procurement leaders and contractors: The Air Force has signaled it will quantify success by a negotiated “measurable amount” reduction in manual compilation, creating a performance-driven procurement metric; how that metric will align with existing contracts — such as the $31 million award to Air — is not specified in the notice.
  • The general public and those tracking nuclear readiness: The ask reinforces that the Minuteman III fleet — 400 missiles kept ready as a “no-fail” mission — must be sustained longer than planned because Sentinel faces years of delays, a practical pressure that helps explain why the Air Force is pursuing new digital tools now.

Responses to the RFI were due today. The notice leaves a clear immediate question on the record: how will new AI capabilities, existing contracts such as the $31 million award to Air, and the Air Force’s recent experimental work with autonomous agents fit together to deliver a single, trustworthy data picture for a mission the notice calls “no-fail”? The Air Force asked industry to propose a fix; industry has until today’s deadline to respond, and the interactions among those pieces remain to be shown.

Original story at Breaking Defense