TL;DR
- A brag doc is a running, unedited record of your real work — not a polished CV.
- Capture wins and achievements, impact (numbers), challenges and decisions, and feedback and lessons.
- The material already exists, scattered across AI chats, emails/Slack, tickets and PRs, past reviews, and old CVs.
- Use AI to find and extract these moments — never to invent or exaggerate them — then verify against your own memory.
- Follow five steps: collect, extract, verify, organize, maintain.
- Maintain it in small increments, a few minutes every week or two, instead of rebuilding it once a year from memory.
- It becomes the source everything else — your CV, reviews, promotion cases, interview stories — gets pulled from later.
Think back six months. You solved something hard, made a call nobody else wanted to make, or quietly fixed a mess before it became anyone's problem. Could you describe it right now, with a date and a real outcome? Most people can't. Not because the work wasn't good — because nothing captured it at the time, and memory alone doesn't hold detail for long.
A brag doc is the fix. It's a running, unfiltered log of what you actually did at work, kept for yourself, added to as you go — so that when someone finally asks "what have you accomplished," you're reading from a record instead of reconstructing one from scratch under pressure.
A brag doc isn't a CV
A CV is a highlight reel, edited for a specific reader — two pages, curated, polished, missing everything that didn't make the cut. A brag doc is the opposite: it's everything, unedited, written for an audience of one. It's the messy source material a CV gets built from later, along with your performance review, your promotion packet, and whatever story you tell in an interview six months from now.
That distinction matters because it changes what belongs in it. A brag doc should hold more than your CV ever could:
- Wins and achievements — things you shipped, fixed, launched, or got recognized for
- Impact — anything with a number attached: time saved, revenue moved, adoption, errors reduced
- Challenges and decisions — hard calls you made, tradeoffs you defended, problems you untangled
- Feedback and lessons — praise you received, mistakes you learned from, things you'd do differently next time
Nothing here needs to sound impressive yet. That's what editing is for, later, when you actually need a version of it for someone else to read.
The material already exists — you just haven't collected it
The instinct is to sit down and try to remember. Don't. Go looking instead. Most of what belongs in a brag doc is already written down somewhere, just scattered:
- Your AI chat history — ChatGPT, Copilot, Gemini, Claude — anywhere you've described a real problem while solving it
- Old emails and Slack messages, especially ones where someone said thanks, asked how you pulled something off, or you explained a decision
- Jira, Linear, or GitHub — tickets closed, PRs merged, commit messages, code review threads
- Past performance reviews, 1:1 notes, and any written feedback you've saved
- Old CVs, resumes, and presentations you've already written — they're full of things you once thought were worth including
Collecting from five sources sounds slower than writing from memory. It isn't — it's faster, because you're not inventing anything, only finding what's already there.
Use AI to find it, not to make it up
This is the one rule that matters more than any other: AI's job here is retrieval, not invention. It's very good at reading through a pile of your own words and surfacing the moments worth keeping. It should never be asked to guess at an outcome, round a number up, or make something sound bigger than it was. A brag doc full of things you can't actually back up is worse than no brag doc at all — it fails you exactly when you need it most, in a review or an interview where someone asks a follow-up question.
Copy this prompt
"I want to build a brag doc — a running record of my real work, not a résumé. Go through what I've shared with you and pull out everything that could belong in it:
— Wins and achievements: things I shipped, fixed, launched, won, or got recognized for
— Impact: anything with a number attached — time saved, revenue, users, errors reduced, adoption
— Challenges and decisions: hard calls I made, tradeoffs I defended, problems I untangled
— Feedback and lessons: praise I received, mistakes I learned from, things I'd do differently
For each one, tell me roughly when it happened, what the situation was, what I did, and what resulted. Only include things you can actually point to in what I've shared — don't infer, guess, or round anything up. If something's unclear or you're not sure it counts, flag it as uncertain instead of stating it as fact. I'll verify and organize everything myself afterward."
The process: collect, extract, verify, organize, maintain
- Collect — pull the raw material together from wherever it lives: chats, tickets, reviews, old documents. Don't filter yet, just gather.
- Extract — run the prompt above against each source, and let AI do the reading you don't have time to do yourself.
- Verify — check what comes back against your own memory. Fix anything that's off, drop anything you can't actually stand behind, and keep the numbers honest.
- Organize — group it however makes sense to you: by quarter, by project, by role. The structure matters far less than having everything in one place.
- Maintain — the step everyone skips. More on that below.
Don't build it once a year — keep adding to it
The version of a brag doc most people attempt is a frantic one-time effort the week before a review, trying to remember twelve months in an evening. It never works well, because by then most of the detail is already gone. The version that actually works is boring: a few minutes every week or two, right after something happens, while it's still sharp. That's the entire difference between a brag doc that's useful and one that's a guess dressed up as a memory.
The payoff
None of this is about writing better sentences. It's about never starting from zero again. Once the raw material exists, every version of "tell me about your work" gets easier — the CV you tailor for a specific role, the self-review you write in twenty minutes instead of two hours, the promotion case you build a real argument for, the interview story you tell with actual specifics instead of a vague summary. The brag doc is the source. Everything else is just an edit of it.
Frequently asked questions
What is a brag doc?
A running, unedited record of your real work — wins, achievements, impact, challenges, decisions, feedback, and lessons — kept for yourself and added to over time, rather than reconstructed from memory when you need it.
How is a brag doc different from a resume or CV?
A CV is a curated, polished summary written for someone else to read. A brag doc is the opposite — everything, unfiltered, written for you. It's the raw material a CV, a review, or an interview story gets built from later.
Can AI write my brag doc for me?
AI should find and extract things you've actually done from your own chats, tickets, and documents — not invent or exaggerate them. Always verify what it surfaces against your own memory before treating it as fact.
How often should I update my brag doc?
Regularly — a few minutes every week or two, right after something happens. A brag doc built once a year, from memory, misses most of the detail that makes it useful.
Related reading: How to prep your work history for bulk import, for turning everything you collect here into individual, dated entries.
Capture it as you go, and it's already there when you need it.