Building your profile
Everything downstream reads from one structured profile, so the only step that really matters is getting that right. There are three ways in, because people arrive holding different things, and all three land on the same editable draft.
Upload a resume
A PDF or a document you already have. I parse it into roles, dates, highlights, education, projects, and skills. Costs 2 credits. Pasting the text instead of uploading the file works identically and costs the same.
Hand over public links
A profile URL, a personal site, a GitHub handle, or just your name and current employer. I search the open web, read what I can reach, and build the profile from that, then show you the list of pages it actually came from so you can see where each claim originated. Costs 8 credits, because it is live web searching plus two model passes rather than a single parse.
Type it yourself
Free, and always available. The same editors used to correct a parsed profile work from empty.
Nothing is invented
The research pass is the only step that could put something in your profile you did not write, so it is weighted hard toward “could not find it” over guessing. A confidently wrong profile is worse than an empty one: you have to notice it and undo it, and every resume written before you notice is fiction.
That is also why the draft is always shown back for correction before anything is written from it. Edit it then, or any time after.
Target role and goals
The last step of intake asks what you are looking for: target role, seniority, locations, and whether you want remote work. That is what the job feed ranks against, so a profile without it still writes resumes but produces weak matches.
Profile completeness
A completeness figure is shown as you fill it in. It is not a score you are being graded on; it exists because a thin profile produces thin output, and it is better to be told that up front than after twenty applications.