Every recruiter deserves a Twin.

It reads every candidate against your standard, shows the evidence for every verdict, and remembers why you decided. You make the calls.

you approve · undoable for 72 h · every step cited

GreenhouseGreenhouseRésuméGmailGmailScreeningInterview notesLinkedInLinkedInWorkableWorkable
Recruiter Twinworking
MCMaya ChenBackend · 8 yrs · Berlin46confidence
Reading résumémaya-chen.pdf
Extracting evidence
Comparing to Scorecard v3
Updating confidence
Learning from 214 decisions
Recommendation
readingscoringremembering
Hiring memory updated+1 fact · 132 total
Synced to GreenhouseMaya Chen → Interview
Interview scheduledThu 14:00 · hold placed

Hiring software remembers everything.
Except the judgment.

Your ATS knows who applied, when, and what stage they reached. It has no idea what good looks like for your team, what the evidence said, or why the call was made. Twynit holds exactly that — and the Recruiter Twin is how it works beside you.

How a decision happens

One standard. Applied to everyone. Shown to you.

01 / 04

You write the standard once.

Describe your ideal candidate in plain English. It becomes a Scorecard — weights, mandatory signals, screening questions — that you can read, edit, and version like a document.

scorecard · senior backend engineer · v3

“Someone who has owned production incidents end-to-end. Kafka is non-negotiable. I care more about reliability instincts than titles.”

becomes

Skills
35
Experience
25
Qualitative fit
20
Location
10
Education
10
Kafka — mandatoryIncident ownership — mandatory+ 2 screening questions written

02 / 04

Every résumé is read against it.

Not skimmed — read. Every candidate lands ranked with the reasons written down, whether ten résumés arrive or a thousand. The standard never gets tired at 5 pm.

pipeline · 34 read · ranked by your scorecard
Advance
Maya ChenAll 3 mandatory signals, cited
J. OseiIncident ownership at scale
Consider
Daniel OkaforStrong platform work, domain unproven
R. PatelKafka adjacent — Pulsar in production
Hold
Priya Nair2 signals unestablished → questions sent

every rank has a reason · nothing is a black box

03 / 04

Every verdict shows its work.

Evidence cited to its source. The case against, argued. Confidence stated honestly. When something can't be established from the résumé, it becomes a screening question — never a guess.

match breakdown · maya chen

Maya Chen

Backend · 8 yrs · Berlin

Advance

Owned incident response for a 40-service payments platform

résumé · p.1

Kafka in production for 4 years — your mandatory signal

résumé · p.1

Team-lead experience — not found

gap · noted

The case against: no lead experience. Your scorecard weights it low — stated, not hidden.

high confidence · every line above links to its source

04 / 04

You decide. It remembers.

Approve, override, or hold — every call enters a permanent ledger. Override your Twin and it asks what it missed. Your answer changes how it reads from then on.

decision ledger · this search

#48 · Maya Chen → Interview · by you · evidence attached

#47 · J. Osei → Interview · by you

#46 · T. Brandt → Pass · override

You overrode the verdict. What did your Twin miss?

“Payments domain matters more than raw scale for this team.”

Remembered — applied to how the next hundred are read

Memory

What your company learns, it keeps.

Hiring knowledge usually walks out the door with the people who held it. In Twynit it compounds: decisions become outcomes, outcomes become facts, facts propose changes to your Scorecard — and you approve every one.

The ledger

#48 · Maya Chen → Interview · evidence attached

#41 · A. Fontaine → Hired · day 90: exceeding

#33 · K. Yamada → Hired · day 180: promoted

… 214 decisions, every one with its why

A fact your company now owns

Incident-response ownership predicts success in your backend hires better than title seniority.

12 outcomes · 3 searches · high confidence · evidence attached

The scorecard evolves

proposed from outcomes

+ Incident ownership · weight 15 → 20

Title seniority · weight 10 → 5

Approved by you → Scorecard v4 · pipeline re-read

And the judgment is audited like any asset: every verdict is checked against what actually happened. Your Twin tells you when it was wrong — and what it changed because of it.

The stack

Your ATS stays the system of record.

Twynit doesn't replace your stack — it does the judgment work on top of it. Candidates flow in from the systems you already run; every decision writes back with a receipt that names exactly what changed, where.

Greenhouse
Lever
Ashby
Workday
Workable
BambooHR
Zoho Recruit
Gmail
Google Calendar
Microsoft Outlook
Slack
Microsoft Teams
writeback · live
just now

Posted to #eng-hiring — shortlist ready for review

Slack
1m ago

Interview hold placed — Thursday 14:00

Google Calendar
3m ago

Screening invite sent — expires in 7 days

Gmail
6m ago

Updated Greenhouse — Maya Chen → Interview

Greenhouse

every write is logged · attributable · undoable for 72 h

Trust

It asks before it acts. Until you say stop asking.

Every action starts in Ask mode — previewed, and done only when you approve. Approve the same action five times unchanged and your Twin offers to take it on. In Announce mode it acts and leaves a receipt, reversible for 72 hours.

Every action is previewed — or receipted

72-hour undo on anything announced

Every decision lives in a permanent, auditable ledger

Send screening invites to 3 candidates?

All three cleared your threshold with Kafka established. Invites expire in 7 days.

Dismiss

Try both. The switch is real.

Judgment compounds — when something remembers it.

Write your standard this afternoon. Your Twin reads everything from day one.