Ultra PatternsAI interface patterns with receiptsULTRA project

Pattern 07 · Dashboard with AI insights, Chat or assistant · Explanation pattern

View

Why is this urgent

Seen in a case Built for a real case and observed with its users, or built for a case whose user test is still to come. No controlled comparison yet.

Practice view. What to build, the words on screen and how to measure it. Switch to Research for why it should work, the framework properties and the references.Research view. The research side is shown below the card: why it should work, the properties it targets and the references. Switch to Practice to hide it.

Next to an AI judgement such as urgency or category, a small "why" reveals one sentence quoted or derived from the source that justifies it.

WhyAlways-on explanations add noise to a list; explanations buried two clicks deep are rarely opened. One line on demand gives people a way to check the judgement without slowing the scan.
WhenLists and cards where an AI judgement drives what people do next.
Not whenHigh-stakes decisions that need the full reasoning up front; there use a fuller explanation, not a one-liner.

The system shows how it worksCase: CNEXT Mail2Flow (company approval pending)

Done right, done wrong

Schematic screens, not real products. Click a screen to enlarge it.
Right. A small 'why' next to the tag opens one quoted sentence from the e-mail.
Wrong. A paragraph of model reasoning under every item in the list.

In practice

The pattern was drawn for this case’s design prototype, where opening an e-mail shows why it was classified as urgent, the original e-mail and the actions. The prototype tested in July did not carry it, and the test showed where the one-liner is missing rather than whether it is read: the AI’s proposed actions came without a reason, several participants questioned the AI’s logic, and one asked outright why it proposed what it proposed. The report’s recommendation is a ‘Why this suggestion?’ line under each proposed action, which is this pattern applied to actions instead of labels.

Words on screen

"Why high?""Because the e-mail says: 'we need this before the audit on Friday'""Not a good reason? Lower the urgency"

Obligations to keep in mind (AI Act, GDPR)

  • Make the reason checkable against the source so people can overrule it. (O2, IUI 2026, Table 1)

Design considerations, not legal advice or a conformity assessment. What O1 to O5 mean.

Build it

Illustrative signatures, not a published library: the names below are suggestions to adapt to your stack. Thresholds are examples to tune on your own data.

States: collapsed, open, disputed

Component: <WhyThis reason={reason} quote={quote} onDisagree={fn} />

Prompt snippet: Return the single sentence from the input that best justifies the judgement, quoted verbatim, and a five-word paraphrase.

Measure it

Users get how it decides · 10 minMental effort · 5 minPlan your testTen minutes per participant is the usual budget: pick the two or three cards that fit.

Same idea, other names

Try it

Open the demo

Research side · why it should work

Why it should work

The mechanism, notes and references for this card are written together with the science editor at KU Leuven Augment. Until then this side lists the properties the pattern targets and the contexts in which it was seen.

Properties targeted

Understanding model behaviour · Trust · Cognitive load · Axes: scope: local, type: rationale, format: text, interaction: on demand · Vocabulary bridge

Across contexts

ContextUsersStateResult
CNEXT Mail2Flow, Kairos prototypeaccountants, desktopSeen in a casedrawn for the case's design prototype; not in the July test, where several participants questioned the AI's logic and one asked why it proposed what it proposed
Your contextBring your case

Editors

Practice: Abasin Abdul Khalil · Science: Ivania Donoso-Guzmán · card v1.0 · 2026-09-08