--- title: "How to use AI for EPB, NCOER, and FITREP bullets (safely)" description: "How to use ChatGPT or GenAI.mil to draft EPB statements, NCOER bullets, or FITREP comments from your own notes without pasting PII, CUI, or unit details." canonical: https://aiexplainedplainly.com/guides/ai-for-epb-ncoer-fitrep-bullets/ date: "2026-10-22" source: "AI Explained Plainly" --- # How to use AI for EPB, NCOER, and FITREP bullets (safely) *Guide, October 22, 2026* ## TL;DR - **What you get:** an OPSEC-first way to turn scrubbed notes into draft EPB, NCOER, or FITREP wording, with two copy-paste prompts. - **Time:** about 20 minutes for one report once your notes are scrubbed. That's an estimate; we haven't timed it. - **Do this:** strip names, IDs, unit details, and anything operational from your notes before any AI tool sees them. AI can help with the wording on an evaluation. It can't help with the facts, and it's the wrong place for most of what an evaluation is built from. So this guide starts with what stays out. ## What never goes into a chatbot Take these out before you paste anything, into any AI tool: - Names, including yours and the rating chain's. Use [NAME]. - SSNs, DoD ID numbers, dates of birth, and contact info - Unit names, ship or squadron names, locations, and deployment details - Anything classified, and anything that's CUI (Controlled Unclassified Information) - Mission names, readiness numbers, capabilities, and vulnerabilities - Medical, family, counseling, or disciplinary details One people miss: the evaluation itself. The Air Force instruction says evaluations are "Controlled Unclassified Information" forms. Don't paste a draft or finished report into a personal ChatGPT, Gemini, or Claude account. Work from your own scrubbed notes. ## Pick the right tool Evaluations are official work. If your command gives you GenAI.mil, use it. The Department of the Navy's January 2026 memo calls it "fully authorized to handle CUI," but also says PII and PHI "are not permitted for use on this platform." So the scrub step above applies there too. If your command has approved a different tool, or no tool at all, its policy wins. Personal accounts? See [Can I use ChatGPT in the military?](https://aiexplainedplainly.com/guides/can-i-use-chatgpt-in-the-military/) ## Know which format your service wants The three reports use different styles, and AI won't know yours unless you tell it. | Report | What the official guidance says | |---|---| | Air Force EPB | Uses performance statements, not bullets. AFI 36-2406 defines one as "a standalone sentence" with the action taken and its "impact, results, or outcome." | | Army NCOER | DA PAM 623-3: bullets start with action words, use past tense, and run two lines max, preferably one. DA Form 2166-9-3 uses narrative instead. | | Navy FITREP and EVAL | BUPERSINST 1610.10H says "Bullet style is preferred," and asks for "verifiable accomplishments that are measurable." | Limits depend on the form and block, so I'm not listing them. Get the current rules from your admin shop and paste them into Prompt 2. Marines, Guardians, and Coast Guard members: your service's own guidance applies, and the safety steps are the same. ## Write notes AI can actually use Good input is a plain list of what happened, with rough numbers, written so a stranger couldn't tell who or where. A made-up example: "Took over a stalled tool inventory. Built a sign-out log. Missing items went from about 30 to 4 in two quarters. Trained 3 junior members to run it." If a number is sensitive, describe the change instead: "cut turnaround time roughly in half." Who drafts matters. The Air Force instruction encourages ratee inputs on accomplishments but says the ratee "will not be directed to write or draft any portion of his or her own performance report." If you're the ratee, AI can tighten your input list. The report belongs to the rating chain. ## Prompt 1: draft from your notes ``` You're helping me draft military evaluation wording from my own notes. Format: [EPB performance statements / NCOER bullets / FITREP comments]. Rules: - Use ONLY the facts in my notes. Never add numbers, results, awards, or scope that aren't there. - If a line needs a result I didn't give, write [NEED RESULT] instead of guessing. - Past tense. Start with the action, then the result. - Plain, everyday language. No buzzwords. - No acronyms unless they're in my notes. - Give me 2 versions of each line. NOTES: [paste your scrubbed notes here] ``` The [NEED RESULT] rule matters most. Ask a chatbot for impact and it'll invent some. On an evaluation, that's a false statement in an official record. ## Prompt 2: fit it to the block ``` Here are the formatting rules for this block, copied from my service's current instruction or form: [paste the rules, no names or unit info] Rewrite the lines above to fit these rules exactly. Count characters or lines if there's a limit. If something has to be cut to fit, tell me what you cut. ``` If the AI's output and the instruction disagree, the instruction is right. ## Check every line before it goes up Accuracy is on the people who sign. AR 623-3 says rating officials will prepare reports that are "forthright, accurate, and as complete as possible." The Air Force instruction encourages ratees to check for "inaccurate data" before signing. The Army CIO's AI guidance says users are responsible for products made with AI's help and "should distrust and verify all outputs prior to use." Read each line against your notes: 1. Did this happen exactly as written? 2. Did every number come from you, not the AI? 3. Did the AI add anything that now sounds like a unit, place, or mission? 4. Does it match the box checks or trait marks? ## Finish it in the real form Add names and unit details yourself, in the real form, outside the AI tool. The Army CIO guidance also says to label documents made in whole or in part with AI output; ask what your command expects. Writing awards too? The same method works for [award bullets](https://aiexplainedplainly.com/guides/ai-award-bullets/), and your scrubbed notes can later feed a [civilian resume](https://aiexplainedplainly.com/guides/ai-military-to-civilian-resume/). Start by scrubbing this year's notes today. ## Sources All checked October 22, 2026. - Department of the Air Force, [AFI 36-2406, Officer and Enlisted Evaluation Systems](https://static.e-publishing.af.mil/production/1/af_a1/publication/afi36-2406/afi36-2406.pdf) (Aug. 22, 2025, with AFGM2026-01, Aug. 7, 2026) - Department of the Army, [AR 623-3, Evaluation Reporting System](https://armypubs.army.mil/epubs/DR_pubs/DR_a/ARN43117-AR_623-3-001-WEB-2.pdf) (Feb. 14, 2025) - Department of the Army, [DA PAM 623-3, Evaluation Reporting System](https://armypubs.army.mil/epubs/DR_pubs/DR_a/ARN43006-PAM_623-3-000-WEB-1.pdf) (Feb. 12, 2025) - Navy Personnel Command, [BUPERSINST 1610.10H, Navy Performance Evaluation System](https://www.mynavyhr.navy.mil/Portals/55/Reference/Instructions/BUPERS/BUPERSINST%201610.10.pdf) (July 30, 2025, with changes; CH-3 dated Oct. 1, 2026) - [ASN(RD&A) and DON CIO joint memo designating GenAI.mil](https://www.doncio.navy.mil/FileHandler.ashx?id=24946), Jan. 28, 2026, and the [DON CIO summary](https://www.doncio.navy.mil/ContentView.aspx?id=20456), Feb. 24, 2026 - [Army CIO guidance on generative AI and large language models (ADS-GOV-AI-024)](https://armypubs.army.mil/epubs/DR_pubs/DR_a/ARN41285-PPM_CIO-024-000-WEB-1.pdf), June 27, 2024 *Policies and forms change, so confirm with your rating chain and admin shop. How this is made: an AI editor drafts and fact-checks each article. If we get something wrong, we fix it and note the change.*