Governance
A policy that cannot reach the assessment is not yet a practice.
Akanya is designed to carry an institution’s position on responsible AI use into the place where the work is actually done: the task itself.
Principles the platform is being built against.
The lecturer decides
The lecturer controls assessment requirements, reviews the draft questions and approves the paper. Akanya does not set an institution's academic standards.
In practice: Review and edit generated questions before approving an assessment. Drafting assistance does not remove the approval step.
Assurance by design, not by observation
Confidence comes from how the task is built, not from watching the student. No cameras, no biometrics, no additional hardware.
In practice: Review the student's initial answer, response to a challenge and final judgement rather than a webcam recording.
Proportionality
Deeper reasoning is applied where the consequence of a weak judgement is highest. A weekly tutorial does not carry the load of a capstone.
In practice: Agree the learning outcomes, question demands and time available before testing an assessment with participants.
Evidence a lecturer can point to
A reasoning record is meant to support academic judgement in a review, not to replace it. A person makes the decision.
In practice: Use Review to inspect a participant's reasoning journey. Where Marking Assist is used, the lecturer reviews its suggestion and makes the final marking decision.
Accessibility is not optional
Low-bandwidth written interaction on the existing device, with configurable accommodations, so assessment adaptation does not depend on what a student can afford.
In practice: Confirm device access, timing and accommodation needs with the institution before any student pilot. Do not assume one setup fits every learner.
Say what is built
Anything illustrative, in development or on the roadmap is labelled as such across this site. Akanya is in active development.
In practice: The public demo uses scripted responses and sample insights. It does not run live AI marking or present the sample cohort as an actual pilot result.
This page describes a design approach, not a legal or regulatory opinion. Akanya has not completed a formal legal, privacy or accreditation review, and nothing here should be read as a compliance guarantee. Institutions remain responsible for their own policy, and specialist input would be appointed where a pilot requires it.
Translating policy into practice
Institutional AI policy becomes repeatable assessment practice.
Institutional AI policies establish the principles. Akanya helps translate those principles into repeatable assessment practice, so lecturers are not left to interpret broad policy statements alone.
Akanya could hold the permitted-use position at the level of the assessment rather than the policy document, so a lecturer sets it once when designing the task and the student sees it where the work happens.
How Akanya could operationalise this. Developing alongside institutional pilots.
The reasoning components are chosen per assessment. A faculty can decide that evidence selection and defended judgement are the capabilities that must be visible, and leave the rest of the task unchanged.
How Akanya could operationalise this. Developing alongside institutional pilots.
A quantitative module may need visible working. A law module may need a counterargument. The same framework configures differently rather than imposing one standard across a whole institution.
How Akanya could operationalise this. Developing alongside institutional pilots.
The core journey is written, low-bandwidth interaction on the existing device, with configurable accommodations. Assurance is raised by design of the task, not by monitoring the student.
How Akanya could operationalise this. Developing alongside institutional pilots.
A recorded reasoning journey gives a lecturer something specific to point to in a review: which evidence was marked decisive, what was asked, and what the student did with the challenge.
How Akanya could operationalise this. Developing alongside institutional pilots.
The level of reasoning visibility is a setting, not a rebuild. As capability shifts, faculties can move an assessment class up or down without redesigning the curriculum around it.
How Akanya could operationalise this. Developing alongside institutional pilots.