The Eight-Word Test: What Free AI NCOER Tools Do With Thin Input

// 01The test

Thin input is the honest test for any bullet tool, because thin input is how too many real NCOERs start. If you kept a running log on your Soldier all year, you are ahead of the game. If you did not, you sit down with a support form due, you remember roughly what your Soldier did, and the first thing you type is something like this:

Completed JRTC rotation, built strong offensive operations plan The full input. Eight words. Achieves. SSG. 13F. Metrics field left empty.

That input contains no numbers, no scope, and no results. It says a rotation happened and a plan got built. What a tool does with those eight words tells you everything about whether you can trust it with your Soldier’s record.

// 02Tool one: the fabrication

The first tool, a free AI bullet generator marketed to Army leaders, returned three bullets in a few seconds. Here they are, unedited:

Coordinated fire support operations throughout joint training rotation, integrating six separate fire missions resulting in zero friendly casualties and 100% target destruction rate. Generated output, tested 2026-08-31

Developed offensive operations plan incorporating multi-platform fire support scheme that synchronized indirect fire with maneuver elements across 40-kilometer operational area. Generated output, tested 2026-08-31

Directed targeting efforts during rotation by acquiring and verifying 18 priority targets, enabling commander to maintain initiative and concentrate effects on decisive points. Generated output, tested 2026-08-31

Read those against the eight words that went in. Every quantified claim came from nowhere:

• Six separate fire missions. Not in the input.
• Zero friendly casualties. Not in the input.
• 100% target destruction rate. Not in the input.
• 40-kilometer operational area. Not in the input.
• 18 priority targets, acquired and verified. Not in the input.

The tool did not embellish the facts it was given. It was given no facts to embellish, so it manufactured them. If you paste those bullets onto a support form and sign it, you are attesting to five quantified claims about a rotation that, as far as the tool knows, may not have included a single fire mission. Your senior rater may have been at that rotation. And a board full of senior NCOs knows exactly what a claimed 100% target destruction rate at JRTC is worth.

// 03The part nobody checks: the bullets do not fit the form

Set the fabrication aside for a moment and just measure the output. Those three bullets run 182, 177, and 176 characters. The bullet block on the NCOER holds two lines of roughly 70 characters each, about 140 characters total once the leading “o” is on. Every one of those bullets is three lines long. None of them will fit in the box they were generated for. They also arrive without the leading “o” and with sentence-style periods, so even the parts that fit would need reformatting by hand.

A bullet tool that does not know the size of the bullet box is not a bullet tool. It is a paragraph generator with a military vocabulary.

// 04Tool two: the refusal

The same eight words went into the free BulletForge NCOER generator, with the rating tier set to Exceeded Standard, the tier those fabricated bullets were implicitly written to. It did not return a bullet. It returned this:

INPUT-TIER MISMATCH: The accomplishment “Completed JRTC rotation, built strong offensive operations plan” describes routine performance of assigned duty without a QUANTIFIABLE RESULT or IMMEDIATE IMPACT required for EXCEEDED (DA PAM 623-3 ¶3-11a(3): “immediate impact on the mission, their Soldiers, the unit”) ... As written, this input supports an honest MET bullet: completing a rotation and building a plan are expected duties of an SSG 13F Team Leader. The word “strong” is subjective and does not constitute a quantified result. Please provide the measurable outcome and immediate impact to write at the EXCEEDED tier. BulletForge generator output, tested 2026-08-31 (elided middle section lists the three kinds of evidence that would justify the tier)

The full response also tells the rater exactly what would change the answer: a measurable outcome of the plan, evidence of initiative beyond the assigned task, and the immediate impact on the mission during the rotation. Give it those facts and it writes the bullet, inside the envelope, at the tier the facts support.

// 05Why this matters on a signed record

An NCOER is a signed document. When a rater signs Part IV, the bullets under each attribute are that rater’s signed assessment of what the NCO did. A tool that invents the metrics is not saving you time; it is drafting fiction for an official record and handing you the pen.

And even when nobody catches the invention, the inflation still costs your Soldiers. Boards read hundreds of evals, and when every SSG in the stack claims a 100% something rate, the numbers lose their power to distinguish anyone. The NCO they were true for is the one who pays for that. The regulation’s tier definitions exist precisely so that Exceeded means something. A generator that hands out Exceeded language for eight words of routine duty is working against the only mechanism that makes your top performer stand out.

This is why BulletForge is built to refuse. Not as a limitation we could not engineer around, but as the product’s core promise: it checks your facts against what DA PAM 623-3 ¶3-11a actually requires for each tier, writes at the tier your facts support, and tells you what evidence would honestly justify more. The Honest NCOER Guide walks through those tier markers in detail if you want to see the standard itself.

// 06Run the test yourself

Do not take our word for any of this. The test takes under a minute and costs nothing:

1. Open any free AI bullet tool, or a general chatbot, and type: Completed JRTC rotation, built strong offensive operations plan. Ask for an Exceeded Standard bullet. Count the numbers in the output that you never provided.
2. Open the BulletForge free NCOER generator, no account needed, and type the same eight words at the same tier.
3. Decide which output you would sign your name to.

Then run it a third time, anywhere you like, with real facts: what your Soldier actually did, with the numbers you actually watched happen. That is the version of AI help worth having, and it is the only kind BulletForge will give you.

← Back to The Field Manual