Skip to content

AI-supported writing.
Evidence for assessment.

Students write with AI support. Educators review the submission, recorded activity and optional reflections against their own criteria.

One workspace. Three connected layers.Explore Friction
Companion Peer

What convinced you this supports your argument?

Reflect with the Companion.

An optional study partner invites students to explain, question, and make an idea their own.

From AI use to assessable evidence

Support the work. Invite reflection. Give educators the context to assess.

The three layers

Writing support and
assessment evidence.

Each layer has a distinct job. Together, they connect the student’s work with AI to a richer evidence base for educators.

01
Layer 1 / Create

Task Assistant

A familiar AI assistant beside the working document. Students explore ideas, ask questions, and develop their work within educator-defined boundaries. In-workspace activity builds the assessment record.

RoleWorkspace
Led byThe student
ConnectsWriting and AI
02
Layer 2 / Reflect

Companion Peer

A thoughtful study partner that invites students to explain and evaluate ideas after meaningful pastes. Reflection stays optional: students can respond when it helps, or keep writing.

RoleStudy partner
PostureReflective
ParticipationOptional
03
Layer 3 / Understand

Friction Report

On submission, Friction brings together recorded activity, saved drafts, exact conversations and document excerpts. A clearly labelled AI review connects evidence to the educator’s criteria for human assessment.

RoleEvidence
PostureCriteria-aligned
OutputReport document
Friction workspace / Illustrative example

The contested duty of care in algorithmic decision-making

In recent years, courts have increasingly grappled with whether the common-law duty of care extends to algorithmic decision-making systems deployed by public authorities.
The standard test articulated in Caparo requires foreseeability, proximity, and considerations of fair, just, and reasonable imposition — a structure that maps awkwardly onto opaque algorithmic systems, which operate at a scale and speed beyond traditional duty-bearers.
As established in case law, AI systems must be held accountable under existing frameworks.
The difficulty, as Turner (2019) identifies, lies in locating the defendant
Layer 01 · Task Assistant● live
draft me 3 paras on duty of care for AI
Here are three paragraphs on the contested duty of care in algorithmic decision-making, drawing on Caparo...
paste ↵
Layer 2 / Companion Peer
Your space to think it throughOptional conversation
Illustrative workspace / Writing and reflection, side by side.Layers 1 + 2
The opportunity

Assess the learning
behind the work.

A polished submission tells only part of the story. Universities need a way to understand how students question, evaluate, and apply AI-generated ideas.

For institutions

A foundation for assessing AI-assisted learning across subjects and faculties.

For educators

Context for feedback and academic judgement, grounded in the recorded work.

For students

A place to use AI thoughtfully and show the contribution behind a submission.

As AI becomes part of academic work, the assessment question changes: how did the student arrive at this answer?

Friction brings the writing process into view. Students work with AI, reflect on their choices, and submit a record that helps educators connect the final work to the learning process.

From brief to submission

Familiar for students.
Meaningful for educators.

Educators set the direction. Students do the work. The three layers connect both experiences through a shared assessment record.

A · Student flow

01Creates

Work with the Task Assistant

Draft a response, explore ideas, and ask AI for support in the same workspace.

02Reflects

Think it through with the Companion

An optional conversation invites the student to explain how a passage connects to their own argument.

03Chooses

Student decides whether to respond

Keep writing, reply when useful, or skip. Conversations are recorded; skipping is not evidence of misconduct.

04Submits

Submit the work with its context

Layer 3 brings the recorded process together in the Friction Report.

B · Educator flow

01Designs

Set the assessment and AI boundaries

Define the brief, select criteria, and decide whether to enable optional Companion reflection.

02Receives

Review the Friction Report

Read the submission alongside recorded activity, saved drafts, and exact conversations.

03Reads

Evidence is mapped to rubric

Document and process criteria use separate evidence. Each AI interpretation links to the sources a human can review.

04Acts

Feedback grounded in evidence

Use the submission and recorded sources to discuss the student's work. Academic judgement remains with the educator.

Layer 3 / The Friction Report

More context.
Better-informed judgement.

A readable, criteria-aligned account of the recorded work, with source evidence that educators can inspect. Academic judgement stays with the educator.

The Friction Report arrives with the submission. It doesn’t replace the essay — it sits beside it, surfacing the process the essay conceals.

Clear section navigation, readable evidence and exact transcripts support review on screen or in print. AI interpretations are labelled and linked to their sources; they are not grades or authorship verdicts.

  • 01Concise, clearly labelled AI review
  • 02Criteria-aligned evidence panel
  • 03Recorded session time and activity counts
  • 04Exact optional Companion conversations
  • 05Paste evidence and saved draft changes
  • 06Linked sources and explicit evidence limits
Friction Report · Illustrative exampleLAW 5104 — A. Reed

Friction] Report

A criteria-aligned review with linked evidence.

Review overview

Three sessions and seven paste events were recorded. One passage matched a Task Assistant response; other paste origins could not be established. Review the recorded excerpts alongside the submission before drawing conclusions.

[ evidence ]
Criteria interpretations link to recorded events and submitted passages. They are not a grade or an authorship verdict.
Session time
14min
Paste events
7
Task Assistant prompts
5
Companion dialogues
2
See Friction in action

From an idea
to a working platform.

Friction won the challenge track at the 2026 EduX Oceania Hackathon — designed, built, and deployed in five days. The demo below was the winning submission.

Fig. 02   Live walkthrough · 2026 EduX Oceania Hackathon submission.Challenge track · winner
Questions, answered
Layer 1, the Task Assistant, supports student writing and exploration with AI. Layer 2, the Companion Peer, offers optional conversations that invite reflection. Layer 3, the Friction Report, brings recorded work and dialogue together with a labelled AI review aligned to educator-defined criteria. Together, they connect the learning process to the submitted work.

Plan a class pilot
with Friction.

Tell us about your course, class size and assessment. We can discuss pilot setup, research collaboration or investment.

Talk to the team Explore the platform