> ## Documentation Index > Fetch the complete documentation index at: https://docs.hired.tools/llms.txt > Use this file to discover all available pages before exploring further. # Filling the brain > How to get twenty years of work into a system in an afternoon, what actually belongs in a brain dump, and how to keep it current without it becoming a chore. This is the step people skip and then wonder why the resumes are thin. It is also the only part of Hired that asks anything of you. Everything after this is the product doing work with the material you gave it. ## Start with the job you have now One role, one conversation. Do not open a form. > I've been at Vertex since March 2021 as a senior platform engineer. Let me dump > everything I remember and you file it. Then talk. The assistant creates the role and puts what you said into the brain dump. It does not need it in order, and it does not need it to be well written. ## What actually belongs in there The test is: **would this be hard to remember in two years?** If yes, it goes in. Runtimes, costs, headcount, revenue, latency, error rates, percentages. The number is what makes a bullet survive the follow-up question, and it is the first thing you forget. How big the team was, what you owned versus contributed to, who you reported to, how many people used the thing. Names, not categories. "The nightly reconciliation job" is retrievable later. "Data pipelines" is not. The incident, the migration nobody wanted, the thing you argued for and won. These become interview answers, not just bullets. What a manager or a review actually said. You will not remember the wording, and the wording is the useful part. The failed project, the thing you would do differently. Nobody puts this on a resume, and everybody gets asked about it. ### What does not matter Grammar. Structure. Chronology. Repeating yourself. Contradicting yourself. Nobody reads this document — it is searched, and search does not mind a mess. If you already have a resume, paste it in as a starting brain dump. It is a floor, not a ceiling: the resume is the compressed version, and the point of the brain is to hold the things compression threw away. ## Doing the older jobs Twenty minutes each, oldest last. Older roles need less: what you owned, the two or three things you are still proud of, the numbers you can still remember. The recent ones are what get tailored against. If that sounds like a lot, do one job a day for a week. The material compounds — every dump makes every future resume better, and you never do this job twice. ## Turning raw material into bullets Once a role has a dump, mine it: > Mine the Vertex role into highlights. That is the [`mine_brain_dump`](/workflows#mine-a-brain-dump-into-highlights) workflow. It reads the full dump, checks which highlights already exist so it does not duplicate them, extracts every distinct accomplishment as one bullet in resume voice, rates each 1–5 for strength, tags them for retrieval, and saves them in one call. Then it flags the useful thing: anywhere the dump *hints* at impact but does not give a number. Those are the gaps you can still fill from memory, and they are worth ten minutes. ## Keeping it current The reason career records rot is that updating them is a separate task nobody schedules. The fix is to make it a side effect of talking about your week. > Here's what happened this week: shipped the billing migration, the reconciliation job > is down to 20 minutes now. Had a call with the Helios recruiter — team of six, they > want someone to own billing end to end. Told Priya I'd send her my resume by Friday. That is [`log_my_week`](/workflows#log-what-happened-this-week), and it files each piece where it belongs: work about a job appends to that role's brain dump, a clean accomplishment also becomes a highlight, anything about a company you are talking to becomes a timeline entry and maybe a stage change, anything you said you would do becomes a task with a due date, and the rest becomes a note. Ten minutes on a Friday. It is the difference between a brain that is current and one that describes the job you had in 2023. **Append, never overwrite.** When you tell an assistant something new about a job already on file, the correct call is `append_role_brain_dump`. `update_role` with a `brainDump` replaces the whole thing. Install the [`hired` skill](/skills) and this is handled without you having to watch for it. ## Standing rules Anything that must never be got wrong belongs in a guardrail rather than in your memory of what to correct: > Make a standing rule: I was a contributor to the payments migration, not its lead. > Never describe me as having led it. That saves a note with `kind: "GUARDRAIL"`, and every assistant that connects afterwards gets it in its briefing before it can call a single tool. See [standing rules](/concepts/brain#standing-rules). ## The rest of the profile Worth five minutes once: * **Contact block** — name, headline, email, phone, location, and the links you want on a resume. Every document is seeded from this. * **Personal brain dump** — the things that are about you rather than a job. What you want next, what you will not do again, compensation expectations, non-negotiables. It is what a good summary line and a good "why do you want this?" both come out of. * **Education, projects, skills, certifications** — the four supporting collections. Say them and they get filed. * **A photo**, if you want one on documents. [How the one-photo model works](/concepts/brain#the-photo).