AI HR & recruiting agent. Hire without the busywork.
Every applicant scored against the criteria you published, with the reason attached, and every candidate message written and waiting for your send.
What it does
Resume screening
Scores applicants against the must-have and nice-to-have criteria in your job description, and writes the reason next to every score.
Scheduling drafts
Writes the invite, the reschedule and the nudge, so your coordinator spends the morning sending instead of composing.
Onboarding checklists
Turns your process into a day-by-day list for the new hire, with the owner named on each item.
Reference summaries
Turns the answers you collect into a structured summary, with the quotes it drew from kept alongside.
How it works
Screen
Applications arrive through the form and are scored against your criteria.
Shortlist
Each score carries its reason, so a recruiter can overturn it in seconds.
Draft
Invites, reschedules and rejections are written and wait for a send.
Decide
A person picks who moves forward. The machine never makes that call.
0
Hiring decisions the machine is allowed to make
Minutes
From application to a score with its reason
24/7
Applications accepted through your careers form
15h
HR hours handed off weekly
Target we model in pilots
What the first month looks like
Week one: the assistant shadows your existing process and scores the next batch while a recruiter checks every reason it wrote. By week two the scores are trusted enough to sort by, and the scheduling thread arrives ready to send.
By week four the coordination load has moved and your team spends the day on interviews, culture and offers. Candidates hear back sooner, and every score can be read back with the reason attached.
Evidence · AI HR
What runs today
The parts of this role that exist in the platform right now, and the parts a deployment has to build with you. Nothing in between.
Ships today
A careers form on your site posts applications into one record
The endpoint runs in production: it caps the request body before parsing it, absorbs bot submissions without writing anything, and stores the application in Postgres. One that fails validation gets a reason back instead of a silent success, and the form on your own site also emails you the moment an application lands.
Ships today
Screening produces a score with a reason, and a person decides
The role scores against the criteria you publish and writes why. Every message to a candidate is a draft until your coordinator sends it, and the platform has no send path of its own — which also means it cannot reject anybody by itself.
Scoped with you
An ATS connection is built during the deployment
One platform connector ships today, and it is Meta ads. Greenhouse, Lever, Workable and the rest are integrations we build against your account during a deployment. We name what we have built and leave the logo wall to someone else.
The gates behind the drafts — approvals, spend caps, the kill switch — are listed on the product page.
Frequently asked questions
No. It screens, schedules, and summarizes. Final decisions stay with humans.
It scores against the criteria you publish, writes the reason next to each score, and holds no decision power. You can read every score, overturn any of them, and see the criteria that produced them.
None out of the box today. Applications come in through a form that posts to our endpoint, and an ATS connection is scoped and built as part of a deployment.
Writing the criteria down is the work, and it usually takes under an hour. Once they are in, the next batch that arrives through the form is scored on the following run.
Today the assistant drafts the reschedule against the availability you give it and your coordinator sends it. A self-service booking flow needs a calendar connection, which we scope at the start of a deployment.
The assistant writes the rejection in your voice, following the policy you set, and a person sends it. Nothing goes to a candidate without that step.
That is where the arithmetic works hardest: the same criteria are applied to application number three hundred and application number one, and each score arrives with its reason so you can audit the drift instead of trusting that there is none.