Kessel Run Ran a Two-Day AI Course to Teach Its Own Staff How to Build an App

Thirty people, two days, one app

On 29 June 2026, Air Force Life Cycle Management Center public affairs published a news release announcing that roughly 30 Kessel Runners had spent two days in a class. The class taught them how to build a software application from scratch.

Kessel Run was founded in 2017 to teach the Air Force how to build software. Nine years later, the achievement worth a press release was Kessel Run attending a course on it.

The release, verbatim: “Roughly 30 Kessel Runners immersed themselves in hands-on training during a two-day course meant to teach them the intricacies of AI, June 3-4.” Participants “broke into teams based on experience levels,” and the course “enabled individuals with little to no previous experience in using AI tools, to be able to streamline tasks and deploy fully developed applications.”

Little to no previous experience. At the software factory. In 2026.

The students were the management

Now look at who was actually in the room, because the roster is worse than the headline. Not engineers. The release specifies “Airmen and government civilians in leadership and functional positions.” The software factory sent its managers to learn what software is.

There is a reason there were no engineers. In March 2025 Kessel Run announced it was ending mixed government-vendor coding teams, and that the people writing the code would henceforth be vendors. So by June 2026 there is no large bench of government developers left to upskill. What Kessel Run has is a bench of people who supervise developers.

The release states the purpose of the training without being asked: “By understanding AI capabilities, Air Force leadership can better communicate with industry vendors on realistic timelines and understand the nuances of modern software development. This also enables the acquisition workforce to be better equipped to evaluate the timelines proposed by industry partners.”

That is customer training. Kessel Run is not learning to build. It is learning to grade.

Kessel Run put its own velocity on the record

The course instructor, design practice lead Jeff Rodanski, supplied the benchmark: “What we’re doing in two days is developing an app that probably would have taken us four to six months to build, two to three years ago. The pivot is now you can have one person play all those roles because of the power of AI.”

Four to six months. For one app. That is the number Kessel Run’s own public affairs office chose to print, in a release it wrote about itself, under the heading of good news.

Sit with the sequence. An organization founded to end slow government software measured its own speed, found half a year, and issued a photographed press release about it. Nobody leaked this. Nobody filed a request. Kessel Run walked its confession to the podium.

Lt. Col. Camdon Cady, Kessel Run’s Operations Command and Control materiel lead, delivered the sales pitch to the workforce: “These are tools that we can use to actually make our own lives easier.”

Read the pitch again. Not faster for the operator. Not more capability downrange. Easier lives. The closing argument to the workforce of a wartime software organization is that the new tool means you personally have to do less.

The motto painted on the wall was Code. Deploy. Win. The motto in the room is Less Work. Nine years on, the revolution’s pitch to its own staff has become the pitch of every enterprise software vendor it was founded to replace, and it is landing.

The most powerful tool of the decade, assigned to the paperwork

Jennifer McBee, Kessel Run deputy division chief, on what the training is ultimately for: “Ultimately, the goal of the training is to automate and accelerate administrative work so we can dedicate more time to delivering lethal capabilities to our operators.”

Read that again. The stated goal of artificial intelligence at the Air Force’s flagship software organization is administrative work. Not the air tasking order. Not the weapon system it has been failing to replace since 2017. The admin.

Which, in fairness, is correctly targeted. Admin is what Kessel Run produces. It is the output, the inventory and the finished good. Aim the most powerful tool of the decade at the one thing this organization reliably manufactures and you get faster slide decks, tidier timelines, and a briefing about the time saved preparing the briefing. The machine will automate the paperwork, which will free up capacity for more paperwork, which is the business.

McBee also explained what the skill is worth in a negotiation: “When a vendor tells me it’s going to take four months to deliver something, you can look at it and say, ‘I could do that tomorrow using AI and get it done within a day.’”

Take that entirely at face value, because it deserves to be taken at face value. A Kessel Run deputy division chief believes she can do in one day what a vendor quotes at four months. Now look at the schedule Kessel Run published for its flagship program. Next-Generation Air Operations Center: request for information February 2026, industry engagement March, draft request for proposals August 2026, final November 2026, award June 2027.

Sixteen months to pick the vendor whose four months she could allegedly do in one day.

Hold both numbers up together. Kessel Run has, by its own account, acquired a technology that compresses four months into a single day. It has aimed that technology squarely at its own paperwork. And its own paperwork still takes sixteen months. Whatever AI is accelerating at Kessel Run, it is not Kessel Run.

“2027 is not getting any further away”

That is Cady again, in the same release: “2027 is not getting any further away. We need to be able to be efficient and get things done with AI.”

He did not say which 2027. Kessel Run’s calendar offers two candidates. The Next-Gen AOC contract award is scheduled for June 2027. SAIC’s contract to sustain and modernize the Air Operations Center weapon system runs out in 2027.

Both are contract dates. Not a fielding date. Not an operator need. Not the weapon system that has waited nine years. Money.

Nine years of war-fighter urgency, and on the evidence of its own calendar, the only deadline Kessel Run has ever visibly sweated is the one where the funding runs out.

Nineteen words to say where it sits

The release is careful to note that the class “aligns with the Department of the Air Force AI Hiring and Talent Development Plan,” issued by the department’s Chief Data and AI Office and approved in April 2026. Two days of coursework, escalated to strategic alignment, converted into a photo release. The alignment is the deliverable.

The release then explains what Kessel Run actually is: “a division within the Department of the Air Force Portfolio Acquisition Executive for Command, Control, Communications and Battle Management.”

Nineteen words. In 2017 the entire pitch was that Kessel Run would route around the acquisition bureaucracy. In 2026 its own public affairs shop needs nineteen words of org chart to find it inside one, and the acronym it lives under has its own acronym.

The verdict

Not one fact above was dug out of anybody. Every last one comes from an Air Force press release. Kessel Run wrote this down, ran it through public affairs, cleared it, and published it on purpose, with photographs, believing it to be flattering. That is the part we cannot improve on. The organization read this and thought: yes, that is how we would like to be seen.

Thirty people. Two days. An app that its own instructor says used to take four to six months. A stated goal of accelerating administrative work. A training rationale of evaluating other people’s timelines. And although the release promises teams would “actually deploy it live,” it does not name a single application anyone built, because the application was never the deliverable. The release was.

Kessel Run’s founding claim was that the talent and the tooling were already sitting inside the Air Force, and that only the bureaucracy stood in the way. In June it proved exactly half of that, and not the half it was hoping for. The tooling was indeed sitting right there, reachable in two days by anybody who bothered. The talent gap it spent nine years blaming on the institution turned out to be seated in its own leadership, in the beginner group, taking notes.

Nine years ago Kessel Run promised to teach the Air Force how software gets built. This June its leadership sat down at the back of the room, in the beginner group, for the two-day introduction.

Remedial training. For the faculty. Billed to the students.

Sources

The jokes are ours. These are not.

Disclaimer: Satire. Not one fact above is invented. There was never any need. The opinions are ours alone.