Skip to content

APPLICATION REVIEW · AI SCREENING AGAINST YOUR CRITERIA

A person is more than a PDF.

Lantern reads every application, portfolio and reference against the criteria you agreed, scores it, and turns what is missing into an interview question.

PDF · 6 OF 20 KEYWORDS
PDF · 9 OF 20 KEYWORDS
PDF · 11 OF 20 KEYWORDS
Priya Natarajan
Engineering lead · 2-page CV
READ AS A WHOLE
PORTFOLIO
CODE
DEMOS
VIDEO
REFERENCES
BUILT IT THEMSELVES · PROOF · DEMO + REFERENCE
"Hired the first eight engineers and shipped the marketplace v1 in nine months." Demo opened · reference confirmed.
94
Exceptional
1 gap · asked in the interview

TRUSTED BY TEAMS AT

Baker McKenzie
Crypto.com
HKSTP
Hyatt
HAECO
Sino Group
Hang Lung
Gammon
PrimeCredit
Nan Fung
InsForge
WHAT THEY SHARED
PORTFOLIO
CODE
DEMOS
PAPERS
VIDEO
REFERENCES
→
Daniel Okafor
ONE APPLICANT · read against your taste
YOUR TASTE · V2
Built it themselves
PROOF · DEMO
Shipped to production
PROOF · CODE
Led a launch
NO PROOF · ASK NEXT
Still hands-on
PROOF · REFERENCE
81
Strong
3 proven · 1 to ask

01 · HOW IT READS

Connect the proof across their work.

Everything an applicant shares is read together against the criteria you agreed. What they did lands on a criterion, with the source. What is missing becomes a question.

02 · CRITERIA YOU AGREE

Criteria you agree, not keywords.

The criteria you agree at kickoff are what every applicant is read against. Lantern drafts them from the intake call, the hiring manager approves them, and every criterion gets its proof with the source.

Criteria for Head of Engineering
drafted from the intake call · 45 min
V2 · APPROVED 17 OCT
MUST-HAVE
Built the function from zero
Proof: demo, code or reference
MUST-HAVE
Shipped software in a paid role
Six months is enough for this role
RANKS
Still hands-on
Recent commits, design docs, demos
RANKS
Led a launch
Named in a release, a talk or a reference
✓
Approved by Marcus Ho, hiring manager
Use these criteria
Edit
Add a column for React or TypeScript
↑
64 of 100 shown
Meets every must-have ✕
CANDIDATE
OVERALL FIT ↓
REACT OR TYPESCRIPT
NEW
Iris Tsang
Graduate engineer
88
Very strong
Meets
"…backend in TypeScript"
Calvin Hui
Front-end developer
86
Very strong
Partly
"…junior AI dev, front end"
Daniel Lo
Associate engineer
83
Very strong
Not met
Terence Chan
Software engineer I
80
Strong
Meets
"…mostly TypeScript"
60 more rows · sorted by fit
Column added in 6 s · 11 meet · 8 partly

03 · SAY IT

Say it. The table changes.

Need a criterion you did not agree at kickoff? Ask in plain words. A column appears with the evidence in every cell, a filter narrows the table, a rescore reorders it. Nobody is hidden for good.

04 · FROM 100 TO 5

From 100 to 5, one ask at a time.

Import the applicants from your ATS, agree the criteria, then narrow the table ask by ask. Every cut shows who went and why, and the shortlist goes to the hiring manager with the evidence attached.

100 imported from your ATS

64 meet every must-have

9 led something at school

5 shortlisted · 9 minutes

Use these criteria
100 imported
+95
Only those who meet every must-have
64 shown
36 hidden, not deleted
+60
Led something at school
9 shown
NEW COLUMN
+4
Internships matter more than years. Shortlist the top 5
★ 5 shortlisted
Rescored · weights updated
Iris, Daniel, Peggy, Calvin, Terence
Fit 70 to 85 · every column has its evidence · 9 minutes from import
Open anyone to read the CV and the evidence behind each column
Send to the hiring manager

The resume is written with AI, scored by keywords and cut to one page. It cannot tell you what the person can do.

WRITTEN WITH AI

SCORED BY KEYWORDS

COMPRESSED TO A PAGE

So Lantern reads everything else.

CONNECT YOUR ATS

Connect your ATS and keep working in it.

Applicants come in from the ATS you already run. The shortlist goes back with its evidence attached. Drive the review from Slack, the CLI or any AI tool.

Greenhouse

Lever

Ashby

Workday

SmartRecruiters

Workable

Teamtailor

SAP SuccessFactors

iCIMS

Jobvite

BambooHR

Personio

Recruitee

Bullhorn

UKG

Rippling

Breezy HR

JazzHR

Pinpoint

Manatal

HiBob

Avature

Phenom

Eightfold

Two-way sync with the ATS you run. Not on the list? Ask us.

05 · AFTER THE REVIEW

Screen once. The first interview asks what was missing.

The scored shortlist, the criteria you agreed and each applicant's open question go to AI Interviews, Meeting Intelligence and Analytics. Nobody re-reads the application, and nothing is re-entered.

UPDATED · AI INTERVIEWS

Opens with the missing proof.

LANTERN · 03:40

Your application says you led a launch. Which one, and what did you ship yourself?

UPDATED · MEETING INTELLIGENCE

Carries the proof into the debrief.

Interview pack · Daniel Okafor

SENT 9:30

CV

3 proofs

1 gap to ask

UPDATED · ANALYTICS

RESULT

Shows which criterion is hardest to meet.

Built from zero

18%

Shipped in a paid role

64%

Still hands-on

80%

✓

SHORTLIST APPROVED BY A RECRUITER · nothing moves on its own

The hiring loop

Every decision sharpens the next.

The five agents share one set of criteria. What a debrief decides changes what the next search looks for, what the next screen scores and what the next interview asks.

One hiring taste
V2 · UPDATED 20 OCT
Sourcing
Application Review
scores every applicant against the taste
AI Interviews
Meeting Intelligence
Analytics
Taste updated after the hire
The next search starts sharper.
One loop per role. A person approves every change to the taste.

Hiring is the last moat

Your next great hire
may be bad at resumes.

They still get seen. Bring your next role and we will run the review on real applications, against your taste.

A worked example · illustrative

Read the evidence beyond the CV

An applicant submits a CV, a project portfolio and a recorded demo for a product engineering role.

  1. Bring the materials together against the agreed hiring criteria.
  2. Look for what the applicant personally contributed and the evidence supporting it.
  3. Carry unresolved questions into the interview instead of treating missing information as proof.

The team reviews the evidence and decides what to investigate next.

This example explains the workflow; it is not a customer result or a live product output.

FAQ

Questions about application review

What does Application Review consider?

The page brings together applications, portfolios, demos, papers, video and references against your team’s hiring taste. The aim is to understand the evidence across the applicant’s work.

Is a CV enough to assess an applicant?

A CV is one source. Work samples and supporting materials can provide context that a short document leaves out.

What happens when evidence is missing?

Keep the gap visible as an open question for the next stage. Missing evidence should be investigated rather than presented as something the applicant has proven.

Can our criteria change?

Your team owns the hiring taste. Approved changes become shared hiring context so review and the downstream stages use the same standard.