The AI platform for writing and scoring exams and certification tests. Items come from your own standards and sources, assessors get a reasoned suggestion per criterion, and everything is on record in the version history and the audit log.
Developed with assessment experts and teachers
Sign-in, your assessment platform, delivery and exchange: Examplary only takes on the part where it helps.
We also connect with other systems through single sign-on, QTI and our APIs. Send us a message!
What Examplary does for the candidate, the assessor, the exam committee and the organization, following a certification exam on safe work in construction.
CandidatePreparing for the exam
Candidates practice with questions in the style of the exam and get feedback on every answer. They know where they stand before they sit it, and you see which topics still need attention.
1.5×
as much material stays with people after a week when they test themselves instead of rereading it four times.
AssessorFor assessors and examiners
For every answer, the AI makes a suggestion per criterion, with its reasoning and how confident it is. The assessor approves or adjusts. Every candidate is held to the same standard, even across hundreds of answers. A second assessor can comment on any answer, so the four-eyes principle is on record.
+38%
more consistent scoring with Examplary than between assessors marking the same answers.
Exam committeeFor item writers, reviewers and approvers
Items come from your own standards, syllabus or assessment model. Item writers and approvers work together in the same item bank, with labels per learning outcome and a version history of every exam. After an exam, you see how candidates scored per question and per learning outcome.
OrganizationFor the exam office, IT and compliance
The audit log records who did what: editing an item, publishing an exam, changing a role. With roles and permissions per item, you decide who can edit and who can only view. Data stays in the EU, and you choose where AI processing happens.
Every field is assessed in its own way. Examplary has question types to match, from spotting a hazard to a customer conversation.
Select point
Candidates click the spot in a drawing or photo where things go wrong: an unprotected edge, a badly placed ladder or a missing barrier.
All 23 question typesIntegration
Examplary doesn't need to take over the whole process. Use it where it helps, for writing, scoring or practice, and deliver the rest through your own systems.
Export items and tests to any assessment platform that supports QTI, or import your existing items.
Cirrus customers build a draft exam with Examplary without leaving Cirrus.
Build Examplary into your own environment, with a documented API and, on request, embedded sessions.
Develop question types that fit your exams exactly.
Cirrus customers build a draft exam with Examplary without leaving Cirrus.
Cirrus Assessment
“Everything it produces is transparent and editable in Cirrus, which keeps authors' judgement and subject expertise at the centre of the process.”
Together with
An example with your own numbers. Every number can be changed.
less scoring work
180
hours a year
What we commit to, and where it's written down.
A pilot with one exam, before you go further.
Step 1
In a demo we show Examplary with one of your own exams and assessment models.
Step 2
We set up the workspace with your roles, single sign-on and the connections to your assessment platform.
Step 3
You write or import the items for one exam and let the AI suggest scores alongside your assessors. Then we go through the results together.
In a demo we show Examplary with one of your own exams and assessment models.
In the pilot, compare the AI's suggestions with your own assessors' judgments. That way you see for yourselves how consistent it is.
Can't find your question? Email us at hi@examplary.ai.
As a suggestion, not a decision. The AI makes a reasoned suggestion per criterion with its confidence, and the assessor confirms the result. On the Mohler dataset, Examplary scores 38% more consistently than assessors do with each other. A pilot with your own assessors shows how that works out for your exams.
Under the AI Act, AI that evaluates learning outcomes is high-risk, with requirements such as human oversight, explainability and documentation. Examplary is built for that: no result without an assessor, every suggestion explained, and the technical documentation is public.
Yes. Import items and tests as QTI, or from Word or PDF. Then edit them, label them and export them back to your assessment platform.
Into your own workspace, stored encrypted in Frankfurt. During generation, a copy goes to the AI model, only to carry out that job. Nothing is used to train models, and roles and permissions per item decide who sees what.
Yes. Through QTI 2.1 and 3.0 with any platform that supports the standard, through a direct export to Ans, and inside Cirrus. With the REST API you can also build Examplary into your own systems.
Yes. Using the developer documentation, you can build question types that fit your exams exactly and behave like the built-in ones.
Exam bodies use an organization license. It includes single sign-on, the audit log, your own domain and branding, and shared folders, among other things. In a demo we'll work out the numbers for your exams.
We'll show it with one of your own exams, and answer the questions of your compliance and IT teams.
