Do I have to use an AI assistant?
Do I have to use an AI assistant?
No. The web app is complete: you can type brain dumps, edit resumes, drag cards, run
the CRM and never connect anything.But the product is designed around the conversation, and the two halves are not equal
in effort. Twenty minutes of talking produces a better brain dump than twenty minutes
of typing into a textarea, and “tailor this to the posting and tell me what I can’t
evidence” has no equivalent button.
Which assistant should I use?
Which assistant should I use?
Any MCP client. Claude, Claude Code, ChatGPT, Cursor, VS Code and Windsurf all have
one-paste setup built into the app, and anything else takes a standard
streamable-http entry — or mcp-remote if it only speaks stdio.Two things differ between clients: whether MCP prompts are rendered, which decides
whether the workflows appear as slash commands, and whether skills can be
installed. The workflows are published as ordinary tools too, so nothing is unreachable
either way.Can my assistant see other people's data?
Can my assistant see other people's data?
No. A connection token resolves to exactly one account, on every request, and every
query filters on it. That is enforced by the type system rather than by a permission
check — a data function that forgot the account id does not compile.Admins are the same: they manage accounts, and there is no path from an admin to
another person’s content.
What stops it inventing things on my resume?
What stops it inventing things on my resume?
Three layers, and they exist because the failure is real.The server briefing tells every connecting client, before it sees a tool, never to
invent experience, employers, dates or metrics. The skills carry the same rule with
the specific failure mode named — quiet upgrading during tailoring, not fabrication
from nothing. And your own standing rules are carried in that briefing, so a
thing you have been burned by is stated before any work starts.None of that is a guarantee. Read the gap list you get back; it is the part that tells
you what the resume could not evidence.
Is my data used to train anything?
Is my data used to train anything?
Not by Hired — it is your database on your server, and nothing leaves it except the
handful of optional outbound calls listed in
security.What your assistant’s vendor does with a conversation is between you and them, and
worth checking. Anything you ask about your career passes through their model.
The app used to have its own Docs page — where did it go?
The app used to have its own Docs page — where did it go?
Here. It listed every tool from the same array these pages are generated from, which
made it a second copy of one thing, and the second copy is the one that goes stale.
/docs on your instance now redirects to this site.Two things it carried did not move, because they could not. The skill files are
served by your instance — byte for byte the copies in the repository it is running —
so they live on Settings → Connections now. And your live tool count is in the
header of that same panel, where Test proves it by calling the endpoint.Can I import my existing resume or LinkedIn export?
Can I import my existing resume or LinkedIn export?
There is no import tool yet — that is the next thing being built. But there is
no need to wait for one, and your assistant already knows what to do.On an empty workspace the briefing tells it to ask whether you have a resume or a
LinkedIn export to paste, take whatever comes back, and file it itself: your profile,
a role per job with the raw material in its brain dump, highlights once there is
something to draw them from. Paste and say “file this”.Treat it as a floor rather than a ceiling. A resume is the compressed version of your
career; the point of the brain is to hold what compression threw away.
Why the Harvard format?
Why the Harvard format?
Because it is the one recruiters have read ten thousand times, and because it has
nothing in it for an applicant tracking system to trip over: one column, real text, no
graphics, no accent colours.It is a starting point, not a cage. Four other templates ship, and the Design menu
changes template, font, size, leading, accent and margins per document.
What happens if I stop paying on the hosted instance?
What happens if I stop paying on the hosted instance?
Your account is suspended, not deleted. Sign-in and assistant access stop; the data
stays. Paying again turns it back on.
How do I get my data out?
How do I get my data out?
On a self-hosted instance,
pg_dump — the database is the whole thing, photos
included, because they are stored in the owning row rather than in a file store.From either kind of instance, get_brain_snapshot returns everything in the knowledge
base in one call, and list_resumes plus get_resume returns every document as JSON.
Ask your assistant to write them to files.What does it cost to self-host?
What does it cost to self-host?
Whatever your host charges for a small app and a Postgres. There is no licence fee, no
account, and no phone-home. AGPL-3.0: self-host it, modify it, run it for yourself and
your friends. The one obligation is that if you run a modified copy as a service for
other people, you publish your modifications.