Remote proctoring has opened up assessment to candidates who could never reach a test centre. But getting someone logged in is not the same as giving them a fair chance. Equity begins after login, and it depends on design choices most assessment organisations can control.
We’re heading to Birmingham for World of Learning on 6–7 October, two days on learning, development, skills and the technology behind them. Underneath a lot of those conversations sits one question: how do you know whether someone can actually do what they’ve learned? Here’s what we’re watching, and where to find us.
The Institute of Legal Finance & Management (ILFM) has chosen Cirrus Assessment for its high-stakes qualification exams. Candidates will work with real data in an embedded Excel environment and can sit mock exams in the same system beforehand, so exam day holds no surprises.
At World of Learning 2026, visitors to the Cirrus Assessment stand can take a test in Cirrus Neo or turn training material into a draft assessment in minutes. The AI does the drafting, and subject experts make the final call.
Remote proctoring is often presented as the candidate-centric option. But removing the journey to an exam centre does not automatically remove the pressures of sitting the assessment. In this fifth article of The Proctoring Question, we look at why candidate experience belongs alongside security and validity, and how the environment we create can change the evidence an assessment produces.
Cirrus Assessment has partnered with Examplary to bring AI-assisted exam authoring into Cirrus. Authors turn existing course materials into a draft exam, then review and approve it as usual, with all content kept in the EU.
Our new Examplary integration turns your existing learning materials into a complete draft assessment inside Cirrus. Authors start from a full set of questions to review rather than a blank page, and every item goes through the review process you already run.
Eight questions to ask before you adopt AI in assessment, and six things you can do to keep human judgement where it belongs. Because the test is not whether AI lets you do something faster. It is whether it lets you do something better.
Everyone wants to know the best way to proctor an exam. There isn’t one. Every qualification carries its own purpose, stakes, candidates and constraints, so the useful question is not which model is best, but which model fits, and whether you can explain why you chose it.
There is no universal definition of secure enough. The answer depends on what the task exposes, what is at stake if the result is wrong, and the scale you are working at. The third article in The Proctoring Question sets out how to match the control to the risk, and what it costs when you get it wrong.
Answer 12 questions across strategy, delivery, design and data and get a clear, personalised breakdown of where your assessment operation is strong and where to focus next.