The Army Went Off the Rack

Somewhere in the Army there's a program office that has spent three years sewing a suit.

Not a real suit, obviously. A software system. But the process is pure bespoke tailoring: measure the customer down to the millimeter (we call this "requirements definition"), argue about the measurements in committee for a year, source custom cloth, hand-stitch every seam, and schedule the first fitting for FY29.

By which point the customer is a different size. The mission changed, the threat changed, the tech stack changed. So we re-measure, and we sew some more.

On July 22, the Army signed out a directive that says, in effect: stop sewing. Army Directive 2026-19 — "Enabling Modern Software Acquisition Practices" — flips the default. Commercial products first. Configured commercial solutions second. And custom development? The directive's phrase is "extremely rare," authorized only when no commercial product can cover a meaningful chunk of the need.

Buy the suit off the rack. Hem the pants. Get dressed and go to work.

Let's be real people: this should not be a revolutionary idea in 2026. Your bank didn't write its own email client. Airlines don't hand-code their own payroll. And yet the government has spent decades treating every software need like it deserved a hand-sewn one-of-one, because — say it with me — "our mission is unique."

Sometimes it is. Nobody's arguing the embedded code in a missile seeker should come off a SaaS subscription.

But an awful lot of what the Department runs on is HR, logistics, finance, scheduling, and workflow. The most commodity software categories on planet Earth. And our record of sewing those ourselves is... not great.

Receipts. The Air Force's Expeditionary Combat Support System — ECSS, for the acronym collectors — was supposed to be a logistics ERP. From 2004 to 2012 the Air Force spent over a billion dollars on it, and per the Senate investigation that followed, canceled it with no operational capability to show for the money. Eight years, six program managers, five PEOs, zero suit. A billion dollars of cloth on the cutting-room floor, in a software category the commercial world was buying off the rack even then.

That's the extreme case, sure. But every practitioner reading this knows a smaller version. The custom portal. The bespoke dashboard. The homegrown system some vendor built in 2011 that now costs more in annual sustainment than the commercial alternative charges in licenses — and that we can't leave, because the incumbent is the only tailor who knows where the seams are.

The interesting question

The interesting question about Directive 2026-19 isn't whether it's right. It's why it needed to exist at all.

Because none of this is new direction. Congress built the Software Acquisition Pathway in Section 800 of the FY20 NDAA, and DoD stood it up in October 2020 with DODI 5000.87 — minimum viable product inside a year of first obligation, no JCIDS, iterate continuously. Congress liked it enough to write the pathways into permanent law at 10 U.S.C. 3603 in December 2024. Then in March 2025 the Secretary of Defense signed a memo making SWP the preferred pathway for software across the whole Department, with CSOs and OTs as the default front door.

That's statute, instruction, and top-cover, all pointing the same way. And in the summer of 2026 the Army still had to write it down again.

Why? Because memos don't hem pants.

The sewing habit doesn't live in policy. It lives in the fitting rooms. The requirements shops that measure for a suit nobody can sew, because every stakeholder gets to add a pocket. The test community that re-inspects every stitch the vendor already inspected. The ATO process that treats each install of the same product like a brand-new garment. The contracting office that knows FAR Part 15 cold and has never run a Commercial Solutions Opening in its life.

None of these people are villains, by the way. Every one of them is optimizing for the grade we've historically handed out: no findings on your watch, no protest sustained, no audit flag in your file. Custom feels controllable. Commercial feels like risk. The incentive structure did the sewing, one CYA stitch at a time.

To its credit, 2026-19 aims at exactly those rooms: automate testing where you can, accept vendor test data, test while you build instead of after, use continuous ATOs and reciprocity so cyber approvals stop repeating themselves, and put products that fit many organizations on enterprise contracts. That's the right list. It's also the hard list — and it's where this either becomes real or becomes a poster.

Practitioner notes while we watch

There's a lot of thread to pull here, and more than one way to hem a cat.

If you're in government: learn the difference between an alteration and a re-weave. Configuring a commercial product — hemming — is the whole deal. A "configured" product that's sixty percent custom code with a store label stitched inside the collar is the old bespoke habit wearing a disguise, and it comes with the same sustainment tail that ate your last budget.

If you're in industry: the CSO is the front door now. Show up with a product, not a proposal to build one. "We could develop that" is a sewing pitch. The Army just told you the sewing era is over.

And if you're grading this a year from now — don't count memos. Count fittings. How many commercial products fielded in months instead of years. How many ATOs honored through reciprocity instead of re-run from scratch. How many CSOs the contracting shops actually executed.

The directive was the easy part.

Suits get won in the fitting room.