Edge AI Screening

AI that reads, asks and listens.
People who decide.

Edge uses AI at every stage of screening, assessment and interviewing: drafting criteria, testing CV claims, interviewing candidates, supporting human interviewers and analysing what was said. Each step has a check on it and a person behind it.

Every place we use AI Book a demo

Simon · AI interviewerSpeaking

Example question

Your CV says you led the CRM migration. Walk me through what you personally owned, and what someone else decided.

Spoken conversation. Simon talks back. The candidate can choose, or the project can fix one mode.

Every place we use AI

For each one: what the AI does, what checks it, who decides.

Choose a stage. Every AI step below follows the same three-part pattern, so nothing the AI produces reaches a decision without being checked and owned.

From a job description to a shortlist you can defend.

Criteria from the job description

Paste or upload the job description and get screening criteria back, each with a concrete rule for what counts as evidence.

AI does

Drafts Knockout, Essential and Desirable criteria from the job description, optionally blended with real occupational data for the role.

Held in check by

Each criterion carries an exact evidence rule, so later checking has something specific to test against.

Person decides

A recruiter edits and approves the criteria before a single CV is scored.

CV adjudication

Every criterion is tested against every CV, with the CV text read as it was written, including scanned documents.

AI does

Two independently framed passes, one strict and one generous but honest, judge each criterion. Where they disagree, the lower state wins.

Held in check by

A claimed match must come with a verbatim quote that the system locates in the CV. No verifiable quote means the claim is not credited. Checks also cover timeline coherence, duplicate CVs, authenticity signals and LinkedIn verification.

Person decides

Flagged items go to a review queue. People make the shortlisting call and can see the evidence ledger behind every state.

From role to interview in minutes

A role-specific competency interview or questionnaire, without a consultant.

Give Edge a job description and it drafts the questions. You stay in control of what is asked, how long it takes and what it focuses on.

Bring the role

Paste the job description or upload a PDF, Word or text file.

Choose the shape

Set a question limit, or a time limit of up to 20 minutes at one question per two minutes. Pick strictly questions, or conversational.

AI drafts, you edit

Competency questions are generated from the role. Edit, add, remove and reorder them, and add a focus such as culture or conflict management.

Simon or a questionnaire

Candidates are interviewed by Simon, or complete a ten-question competency questionnaire built from the same role.

Analysis, not a verdict

Every question and answer is analysed into a report and PDF for the hiring panel, with no score or recommendation.

Human interviewers, supported

Not every interview should be run by AI. So there is Co-Pilot.

For senior roles or sensitive conversations, a person conducts the interview. Co-Pilot sits beside them: it listens, keeps the transcript, tracks what has and has not been covered and suggests what to ask next. It is a separate persona from Simon, and it never addresses the candidate.

Co-Pilot · interviewer viewLive

Live transcript

Interviewer: How did you handle the supplier delay?

Candidate: We re-planned the schedule and I flagged the risk to the client early, but the decision on the alternative supplier was made by procurement.

Analysis and coverage

Clear on their own actions (re-planning, early escalation). The outcome for the client was not stated.

Planned questions covered: 3 of 6.

Suggested follow-ups

What was the result for the client once the plan changed?
What would you have chosen differently from procurement, and why?

Example only. Co-Pilot never speaks to the candidate.

One CV, four versions

Three of these are claims. The fourth is how you find out.

One career, written four ways. Same person, same employer, same twenty-eight months.

Drag across the stage, or use the arrow keys, to move from what was sent, to what was posted, to what was generated — and then to the document that put every claim to him and recorded what came back.

Illustrative candidate and criteria. Edge compares each CV against criteria drawn from the job description, generates questions targeted at the claims that need testing, and puts them to the candidate through a Simon interview in text, audio or video. Simon returns evidence states only — it never scores, ranks or rejects.

Jordan Marsh — screening record

EDGE CV SCREENER
REGIONAL COMMERCIAL MANAGER
SIMON INTERVIEW · VIDEO · 14 MIN

7 criteria · 9 questions put to the candidate Hold — essentials not evidenced

FROM THE CV: 2 EVIDENCED · 4 PARTIAL · 1 NOT EVIDENCED → AFTER SIMON: 2 EVIDENCED · 2 PARTIAL · 3 NOT EVIDENCED

Criterion	CV	After the interview
Leads a team of 10+ ·E	Evidenced	Not evidenced“Four direct. Plus the agency guys, when we used them.”
P&L authority ·E	Evidenced	Not evidenced“I signed to ten grand. Anything over that went to the regional director.”
Regional revenue accountability ·E	Partial	Partial“Eleven, twelve percent? Nine of it was the price rise, that came from group.”
5 years in commercial management ·E	Partial	Partial“Twenty-eight months in the role. Six years in commercial altogether.”
Postgraduate qualification ·D	Partial	Not evidenced“Two modules left. I'll finish it.”
CRM implementation ·D	Partial	EvidencedNamed the vendor, the cutover weekend, and the two workarounds he built when sales data wouldn't migrate.
Category pricing experience ·D	Not evidenced	EvidencedWalked through his own quarterly benchmark. Never mentioned on any of the three CVs.

Evidence states come from the CV and from the interview transcript, each anchored to the line it rests on. Simon returns evidence states only — it does not score, rank or reject.

Two claims collapsed. Two got stronger, including one he never thought to write down. He is not a Head of Commercial with a team of twenty-five — he is a solid regional manager with four reports who quietly runs the pricing. That is still a hire. For a different job.

Jordan Marsh

Visionary Commercial Leader · Transformation Architect · Growth Catalyst

Professional Summary

A dynamic, results-driven commercial leader with a proven track record of leveraging cross-functional synergies to unlock exponential value in today’s fast-paced business landscape. Passionate about empowering high-performing teams to consistently exceed expectations.

Key Achievements

Spearheaded a paradigm-shifting commercial transformation, catalysing a 40% uplift in revenue while simultaneously driving a 40% reduction in operational expenditure.
Orchestrated the seamless end-to-end migration of a mission-critical CRM ecosystem, empowering 25 stakeholders across 12 geographies.
Championed a culture of relentless innovation, mentoring emerging talent and fostering an environment where excellence is not an aspiration — but a baseline.

Core Competencies

Strategic VisioningSynergistic AlignmentStakeholder EmpowermentGrowth AccelerationExecutive PresenceChange Leadership

Committed to driving sustainable, impactful outcomes in an ever-evolving global landscape.

Jordan Marsh

Head of Commercial | Revenue Growth Leader | CRM Transformation | Team Builder | Speaker

Dubai, United Arab Emirates · 500+ connections

#OpenToWork

About

Commercial leader with 3+ years building and scaling high-performing teams across the region. I believe growth is a people problem before it is a numbers problem — and I’ve been lucky enough to prove it. 40% YoY revenue growth, a 25-person org across 12 markets, and a CRM migration delivered on time.

Experience

Head of Commercial — Regional

Meridian Trading · Full-time · 2022 – Present · 3 yrs

Senior Account Manager

Meridian Trading · 2021 – 2022

Top skills

Leadership99+ endorsements

Revenue Growth87 endorsements

Team Management64 endorsements

French — Full professional31 endorsements

JORDAN MARSH

Dubai, UAE · jordan.marsh@example.com · +971 5X XXX XXXX

Personal Statement

A results-focused commercial manager with over 3 years’ experience leading high-performing sales teams and delivering significant revenue growth across the region. Seeking a new challenge in a forward-thinking organisation.

Employment History

Head of Commercial (Regional), Meridian Trading2022 – present

Managed a team of 25 across sales and account management
Grew regional revenue by 40%
Led the CRM migration project across the business
Full P&L responsibility for the region

Senior Account Manager, Meridian Trading2021 – 2022

Managed key accounts and exceeded targets

Education

MBA, Henley Business School
BA (Hons) Business Studies, 2014

Skills

Fluent French · Advanced Excel · Salesforce · Team leadership

References available on request.

the cv they sent

Draft one — the CV

The document Jordan sent. Neat, plausible, and stretched in four places.

Nothing here is quite a lie. One title is borrowed, one date is generous, one team has been counted loosely and one percentage has been rounded up from something that was never measured.

What a screening result looks like

Never a bare score. Every state is anchored to real text.

Each criterion lands in one of four evidence states, and each candidate in one of five outcome bands. The CV Screener page shows this end to end.

Four evidence states

Evidenced
Partial
Claimed, unevidenced
Not addressed

Five outcome bands

Advance
Advance, with probes
Hold: essential partially evidenced
Insufficient evidence on essentials
Does not meet a mandatory requirement

Explore CV Screener

Built to be trusted

The guardrails are part of the product, not a policy document.

AI is only useful in hiring if the people relying on it can explain what it did. These controls are enforced by the system.

Nothing credited without a quote

A claim that cannot be matched to exact text in the CV or transcript is treated as not evidenced, however confident the model sounded.

Disagreement resolves downwards

When the two adjudication passes disagree, the lower state stands. Generosity has to be earned by evidence.

English and Arabic

CV Screener interviews can run entirely in Arabic as well as English, and Simon adapts to the candidate's register.

People clear the queue

Flagged criteria and evidence gaps wait in a review queue for a person. Nothing in the queue is decided by the model.

Monitoring is not scoring

Proctoring and leave-detection protect the integrity of a session. They are never used as behavioural or psychological evidence.

Recordings expire

Interview video carries an expiry date and is deleted when it passes, so recordings are not kept indefinitely by default.

Ready when you are

See AI Screening in a real evidence walkthrough.

Tell us a little about what you're trying to solve and we'll come back with a walkthrough of the evidence model, the adaptive engines, and the reports it produces.

Demo walkthrough

See the candidate journey, assessment modules and reporting flow.

Pilot planning

Map the right assessments to roles, levels and decision points.

AI that reads, asks and listens.People who decide.