Research · Why you can trust it

Can you trust an AI
to judge a candidate?
We asked the toughest judges first.

Siara runs on Integrevise, our proprietary assessment engine. Built with leading assessment experts and proven with real students in universities across Europe and the US, it is now pointed at hiring.

See the evidence
01 · The idea

Most AI screening lets a model decide.
Ours follows a proven method.

Most AI screening asks a language model for its opinion, so results shift from one run to the next. Ours is different. Siara runs on Integrevise, our proprietary assessment engine, built and proven in universities before it ever met a CV. The next few sections show you how it works, who checked it, and what it means for your hiring.

Typical AI screening
Siara
How a score is made
A model's opinion, formed fresh each time
A fixed method, validated by assessment experts
How consistent it is
Varies with the model's mood
One fixed method and scale for every candidate
Why a score was given
A reason written after the fact
The candidate's own words
Who sets the bar
The vendor's model
Your criteria, for every candidate
02 · How it works

One conversation.
Many minds behind it.

Our own AI agents work in concert while the candidate talks. They hand off to one another in the moment, so the candidate simply has a natural conversation.

  1. 01ListensFollows what the candidate actually means, not just the words they use.
  2. 02ProbesKnows when an answer needs more, and asks the follow-up a skilled interviewer would.
  3. 03WeighsHolds every answer against your criteria on the same fixed scale.
  4. 04ExplainsTies each result back to the moment in the conversation that earned it.
03 · What you get

What you get back.
Scores, a shortlist, a brief.

Every score shows its working.

Open any candidate to see how each criterion was scored, the follow-up that was asked, and the words behind the result.

  • Same criteria for every candidate
  • One fixed scale, so results compare fairly
  • Quotes, not guesses

Every applicant, ranked for you.

Nobody gets skipped because the phone went to voicemail. Everyone is interviewed, and the list is ordered against your criteria.

  • 48 interviewed, not 12
  • Ranked against your criteria
  • You decide who moves forward

A brief you can hand over.

One page per candidate, written from the conversation, with room for your own view before it goes to the client or hiring manager.

  • Summary in plain English
  • Attach to your ATS record
  • Your name on it, your call

Reliable

Every answer is scored against the same criteria on the same fixed scale, so candidates are compared like for like.

Explainable

Every score points to the moment in the conversation that earned it. Nothing is hidden inside a model.

Ethical

Every candidate gets the same structure and a fair chance to expand on their answers. Only job-relevant evidence counts, and your team makes the decision.

Professor Stephen DobsonVice-Rector at NTNU, Norway's largest university. Author of Assessing the Viva in Higher Education, the book on spoken exams.View on LinkedIn
04 · Who stands behind it

Validated by a leading expert on
judging people by what they say.

Stephen did his PhD on the spoken exam, then wrote the book on it. He has led university education faculties in three countries and now helps run Norway's largest university. He has validated our scoring method, so every candidate is judged with one fixed method and scale, and every result can be explained.

PhDOn how spoken exams judge what people know, from University College London
3×Head of a university's education faculty, in Australia and New Zealand
100+Books, chapters and articles published over 30 years
05 · Where it was proven

Tested by the toughest judges
of whether someone really knows.

Before recruitment, our method had to satisfy academics whose job is to spot shallow answers. Here is what they found.

Cardiff University
Controlled study · Russell Group

Written tests scored near-perfect. The conversation exposed who really understood.

AI-led viva against human-led viva. Same criteria for every student.
Lindenwood University
Full academic year · Peer-reviewed

8 in 10 students felt more confident in their understanding after explaining it aloud.

Read the paper

Used with real students, for real grades, in more than ten universities across Europe and the US. Then named Startup of the Year 2026 by Digital Universities UK.

In their words

Academics who tried it.

Lecturers and professors from the UK and the US, in their own words.

By prompting students to explain their thinking in a structured, low-pressure environment, it makes their understanding far more visible and highlights where they need to improve in a way that's difficult to achieve through written work alone.

In hiring termsA candidate explaining their thinking out loud shows far more than a CV, and tells you exactly where they fall short.

Laura Jeanne Penrod2024 Nevada State Teacher of the Year
06 · What came of it

A full year in real classrooms.
The time it saves, peer-reviewed.

James championed our year-long pilot at Lindenwood University and is lead author of the peer-reviewed paper that came out of it, which shows how much time the Integrevise method gives back.

250+Publications on AI, learning and technology
PhDIn Artificial Intelligence
1 yearLive pilot with students across departments
Read the paper
Dr James HutsonSenior Professor and Director of AI-Enabled Academic Transformation, Lindenwood University. Lead author of Beyond AI Detection.View on LinkedIn
07 · What it means for you

Lecturers got back 83% of their assessment time.
Your hiring team will too.

30 minper screening call, including prep and notes5 minto read the brief

In the Lindenwood pilot, a 30-minute oral assessment became a 5-minute review. Siara brings the same method to hiring, where a screening call takes about 30 minutes once you count the prep before it and the notes after it, so the maths is identical.

1,667hours back a year
208working days
100%of applicants interviewed
08 · A week of hiring

One role. 200 applicants.
Here's how the week changes.

Press the switch and watch the hours move.

30 min
Mon
1 hr
Tue
1.5 hrs
Wed
0
Thu
0
Fri
3 hrshours spent screening
200 of 200applicants actually interviewed
Wednesdayshortlist ready to share

In today's week, 200 people apply, a quarter get a 20-minute screening call, and the rest of the time goes on chasing and notes. With Siara, every applicant is interviewed on the day they apply and each brief that matters takes 5 minutes to read.

Want your Thursday and Friday back?Try it on your next role
Questions

What hiring teams
ask us next.

Siara is the hiring product. Integrevise is the company behind it and the proprietary assessment engine it runs on, built with leading assessment experts and proven with real students in universities before it was pointed at hiring. Everything on this page about the engine, the research and the people applies directly to Siara, because the same engine scores your candidates.

Both have the same problem, which is that written work no longer proves what someone can do. A lecturer needs to know if a student understands their essay; you need to know if an applicant can handle an angry customer. Same method, same follow-up questioning, same fixed scoring. Only the criteria change, and you set those.

Every candidate is scored with one fixed method and one fixed scale, validated by assessment experts, rather than a language model's opinion on the day. Scores are consistent across candidates and each one points to the words that earned it, so you can explain any ranking and defend it to a client or hiring manager.

Not easily. It's a live spoken conversation, and every follow-up is built on what the candidate has just said. A rehearsed or AI-written answer falls apart at the second question. That's exactly why universities turned to this method when AI broke written assessment.

Every applicant gets the same structure, the same criteria and the same scale, whenever they apply. Only job-relevant evidence counts towards a score, and every score shows the words behind it, so nothing is hidden.

You do. We interview every applicant and hand you a ranked shortlist with the evidence behind each score. Who moves forward is always your call.

From our university assessment model, in which a 30-minute tutor-led oral assessment becomes 5 minutes reviewing the Integrevise result, across 4,000 assessments a year. We apply the same maths to a screening call, which takes about 30 minutes once you count the prep before it and the notes and write-up after it. The figures show staff time, not cash savings.

Yes. The Lindenwood study is published open access in MDPI Trends in Higher Education. Read the paper
Your turn

You've seen the evidence.
Now see it on your own applicants.

Put your next role through the same method, scored against your criteria. 50 interviews on us, no card needed.