ApplyBot

It writes the application. You press send.

ApplyBot finds the roles, tailors your resume from facts you’ve verified, and fills in the form. Every run stops one step short of Submit — so nothing goes out that you haven’t read.

Free while you’re only tracking. No card, no trial countdown.

How an ApplyBot run ends before SubmitVERIFIED FACTS ONLYGREENHOUSE · APPLICATIONAISubmitwaits for youReady to submit14 queued · 0 sent

THE ACTUAL BOTTLENECK

Nobody’s bottleneck is deciding where to apply.

It’s the twenty minutes per posting spent restating the same details into a slightly different form. It’s the tailoring pass that gets skipped at application thirty, because by then you just want it sent.

And it’s the reply that arrives three weeks later referencing a role you can’t immediately place, because the tracker and the inbox were never the same thing.

40
applications in a serious week
× 8 min
retyping the same eight facts
= 5h 20m
not spent on the two that mattered

HOW IT WORKS

Four steps, and the last one is yours.

  1. 01

    Point it at the roles

    Search LinkedIn and Apec from inside the app, or import a CSV of companies and contacts you're already working. Saved postings land in one queue.

    Two sources, one queue
  2. 02

    Verify your facts once

    Confirm the roles, dates, tools and outcomes ApplyBot is allowed to draw on. This is the only material it can ever use — everything downstream is bounded by it.

    Done once, not per application
  3. 03

    It tailors and fills

    For each eligible posting the resume is re-ordered and re-worded against the description, fact-checked by a separate pass, and the application form is filled in.

    Greenhouse and Lever
  4. 04

    You read it and send it

    The run ends at ready-to-submit. You open the finished application in a real browser window, read it, and submit it yourself — or don't.

    Always a human action

GUARDRAILS

The scary part of “AI applies for you” is the AI part.

So none of this rests on a model behaving. Each guardrail below is an ordinary unconditional check in code, and the model only ever operates inside what that check already allowed.

No code path to Submit

An automated run can only ever end at ready-to-submit or held-for-review. Submitting is a separate, manual action in a browser window you're looking at.

It cannot invent experience

Tailoring selects and re-words from facts you verified in advance. A second pass then checks the output against those same facts, not against its own draft.

Two platforms, checked in code

Greenhouse and Lever, matched on the apply URL's own domain. Account-gated systems and Easy Apply are out of scope by design, not on a roadmap.

It stops itself

Three consecutive failures pause the campaign. Daily caps limit both volume and spend, so an unusually large batch of matches can't run away overnight.

WHAT IT ACTUALLY DOES

Four constraints doing most of the work.

RESUME TAILORING

Bounded by what you already confirmed

Your verified facts are the whole universe the tailoring pass can draw from. It re-orders, trims and re-words against the posting; it has nothing else to reach for. A separate pass then re-reads the result against the original facts, so drift in wording gets caught by something that didn't write it.

Tailoring draws only from verified factsVERIFIED FACTSTHIS APPLICATIONreordered, not invented

FORM FILLING

Deterministic first, model only where it has to be

Fields that match a known profile key are filled by exact match, at zero cost and with no model involved. Only genuinely unmatched fields go further — and the model that handles them sees which field keys exist, never their values, so it can name the right one without ever holding your data.

Two-tier form fillingFORM FIELDSexact matchexact matchexact matchmodelsees: field namesnever: your dataanswers with a key

SCOPE

A short allowlist, enforced where it counts

Eligibility is decided by the apply URL's domain, in code, before anything is loaded. Where a posting was discovered is irrelevant — a LinkedIn find that applies through Greenhouse is in scope, and a Workday form is out of it no matter how it was found.

Domain allowlist, enforced in codeAPPLY URLboards.greenhouse.iojobs.lever.comyworkdayjobs.comlinkedin.com/easy-applyallowlistchecked in code,not prompted

REPLIES

Interrupts you for interviews, nothing else

Replies to auto-filled applications get classified as they sync. An interview invitation or a real expression of interest pushes a notification; an automated acknowledgement or a rejection updates the record and stays quiet. Applications you tracked by hand are left alone entirely.

Reply triageREPLIESINTERVIEWAUTO-ACKREJECTIONAUTO-ACKNotificationinterviews onlyeverything elsestays in the queue

PRICING

Cheap, because a job search is temporary.

Full pricing and sprint passes

Free

$0

Track your search, no AI automation

  • Contact & company database (CSV import)
  • LinkedIn + Apec job search
  • Manual application tracking & stages
  • Mail sync (read your replies in one place)
Start free

Starter

$9/month

For a first pass at automation

  • Everything in Free
  • 40 AI-tailored applications / month
  • 1 active campaign
  • Resume fact-checking before every send
Choose Starter

Pro

MOST CHOSEN

$19/month

For an active job search

  • Everything in Starter
  • 150 AI-tailored applications / month
  • 3 active campaigns
  • Reply triage & interview notifications
Choose Pro

Unlimited

$39/month

For an all-out search

  • Everything in Pro
  • Unlimited applications (fair use)
  • Unlimited campaigns
  • Priority support
Choose Unlimited

Every plan includes the AI cost — the caps are what we can actually afford to run, not a teaser. TND pricing is adjusted for local purchasing power rather than converted at the market rate.

QUESTIONS

The ones worth asking first.

Start with the free tier and see the queue fill up.

Tracking, search and your contact database cost nothing. Automation is there when you’re ready for it.