How to use it

Half an hour once, ten minutes a day.

Nothing here is hard. The only way to get a bad result is to rush the knowledge base, so this spends most of its time on that.

The first half hour

Setting it up, once.

  1. 01

    Do the homework before you open the app

    20 minutes, offline

    Open a blank document and write down everything you have done. Every project, every internship, every hackathon, every certification, every position of responsibility. For each one: what the problem was, what you built, which tools you used, and what changed because of it. Numbers wherever you have them, even rough ones. Users, latency, marks, team size, hours saved.

    This is the single thing that decides output quality. The app never invents a fact, by design, so a resume can only be as specific as what you gave it. Three vague lines produce three vague bullets. A page of real detail produces bullets you can defend in an interview.

  2. 02

    Fill in your profile

    1 minute

    Name, email, education and your college. The college is not decoration: alumni are the highest answering group in outreach, and this is the field that finds them.

  3. 03

    Import the document

    5 minutes, mostly waiting

    Upload your resume and that document on the Import page. Your own model reads them and proposes accomplishment chunks, one per thing you did. Read them. Fix the wording. Delete anything wrong. Nothing is saved until you press save.

    Reviewing here is cheaper than finding a wrong bullet on a resume you already sent.

  4. 04

    Set your filters

    2 minutes

    Region first, because it decides which jobs are even stored. Then role families, the most years of experience a posting may ask for, and the cities you would actually move to. Preferred cities are a nudge in ranking, not a wall, because a great role in the wrong city is still worth seeing.

  5. 05

    Fetch jobs

    10 minutes, walk away

    One button. It reads every company you track, drops what does not fit, and scores the rest against your knowledge base in the same run. Go and do something else while it works.

After that

The ten minute loop.

  1. 01

    Open the shortlist

    The home page tells you how many roles are worth applying to today. Start there, not with the full list.

  2. 02

    Read why it scored

    Open a role and the five numbers behind its score are on the page, with the skills you match and the ones you do not. If the reasoning looks wrong, your knowledge base is missing something. Go and add it.

  3. 03

    Generate the resume

    One click per role. Read it, edit any wording that sounds unlike you, download it. It is deleted from the app ten minutes later, so download it before you close the tab.

  4. 04

    Do the outreach

    Every role comes with searches for five kinds of people and a short opening message. Open the alumni one first. Send three. It takes four minutes and it is the highest return part of the whole process.

  5. 05

    Mark it applied

    So tomorrow's shortlist is the roles you have not touched yet.

Every week or so

Keeping it sharp.

  1. Add what you built this week

    Finished a project, shipped a feature, cleared a certification? Put it in the knowledge base the same day. Every future resume gets it for free.

  2. Add companies you care about

    Search any company by name in filters. If it has a public board, one click starts tracking it.

  3. Fetch again

    New postings, expired ones marked, everything rescored against a knowledge base that now knows more about you.

Do not do this

The four ways to get a bad result.

  • Uploading only your resume

    A resume is already compressed to one page. Feed it the long, ugly, detailed version of your career and let the app do the compressing, differently for every role.

  • Skipping the review step

    The model proposes, you approve. Thirty seconds of reading here is what makes every later bullet trustworthy.

  • Leaving the region unset

    It is a hard filter, not a preference. Set it wrong and you will be reading jobs you cannot take.

  • Judging it before you have fed it

    An empty knowledge base scores everything badly and writes nothing worth sending. That is not the app being wrong, it is the app refusing to make things up.

That is the whole thing.

Set it up properly once, spend ten minutes a day, and put back into the knowledge base whatever you build. It gets better at the job the longer you use it, because it knows more about you and nothing about anyone else.

Getting it running

Four commands, once.

About fifteen minutes, and most of that is a download you can walk away from. No database to install and no Docker. You do not need to have anything set up first: the script installs Python, Node and Ollama itself if they are missing.

01
02
03
04

Add yourself to the wall and the setup steps open up.

Join the wall