Sourced AI hides which fact came from which source behind aggregate footnotes.
← Back to patterns
People + AI Guidebook Information Foraging with Generative AI Explainable AI in Chat Interfaces
Trust & Verification
Citation Trail
Source Mapping for Generated Claims
Hover any sentence in an AI answer and the contributing sources highlight in place. Each fact carries a visible thread back to where it came from — no more anonymous synthesis behind a single footnote number.
Metaphor
Footnote Index
Interaction
Hover, Highlight
Use case
Research Tools, Sourced Q&A, Knowledge Bases
Failure mode
Hidden Structure
Control mechanisms
Hover Mapping · Source Pins · Sentence-to-Source Links
Reference examples
Perplexity · Glean · Notion AI Q&A · Granola
Interactive Example
The live side-by-side is the comparison.
This prototype already contains the contrast between chat and the pattern response. It stands as the primary comparison surface for this page.
A footnote number proves nothing. A source span does.
Sourced chat answers collapse a paragraph of synthesis behind a single citation marker. The reader sees a [3] at the end and has to trust that every clause before it actually came from source three. Verification becomes an act of faith rather than something the interface supports.
Citation Trail binds each sentence to the specific source spans that produced it, revealed on hover and inspectable on click. The user can trace any claim to its origin and remove a source they distrust — turning citation from decoration into a working audit surface. Reading and checking become the same gesture.
A citation you cannot follow to a specific span is a claim wearing the costume of evidence.
DOPE Lens
How this pattern improves AI interaction quality.
Observability
Every claim shows its source spans on hover
Directability
Click a citation to inspect or remove its source from the answer
Predictability
The same claim reliably maps to the same set of source spans
Explainability
Each output sentence is bound to specific source spans, not just a footnote number
Fit
Where it earns its place.
Use it
when
when
Readers must verify which source supports which specific claim, not just the document overall.
Avoid
when
when
The output is opinion or synthesis where granular source mapping would be misleading.
In the wild
Where this pattern already ships.
Real implementations from products users already know. Each entry shows how the pattern got applied — what shipped, where it sits in the product, what it traded off.
Perplexity
Answers carry inline source chips named by domain (for example "bmj +1"). Hovering or selecting one opens a card that pages through the sources behind it, with title, snippet and, for domains Perplexity has rated, a Government, Academic or Trusted label. To see the passage behind a sentence, you select it and choose Check sources: a side panel lists the "Sources that support this claim" with the matching words highlighted in each excerpt. That takes a deliberate step; nothing highlights on hover.
Visit Perplexity ↗
Glean
Inline markers follow the statements that need backing, and hovering one previews the source with the cited text highlighted inside its surrounding passage, plus a page number for some sources. That passage-level highlight comes from deep-linked citations, which switch on with Glean's agentic engine and vary by connector and admin settings. Clicking a citation opens the source in its native app; Glean lists going straight to the cited page as not yet supported, though some file types, such as presentations, may open at the cited slide.
Visit Glean ↗
Notion AI Q&A
Notion
Q&A searches the workspace pages you can access, cites specific blocks from those pages in its answers and shows the pages it used; if the workspace holds no answer, it says so rather than give an unsupported one. Connect Slack through Notion AI Connectors and it answers from messages in your Slack workspace too, citing the specific messages it referenced. Block-level citation is finer than a page footnote, but the help page doesn't say whether each sentence carries its own marker or whether a citation previews on hover. Business and Enterprise plans.
Visit Notion ↗
Granola
Beside each point in the AI-enhanced meeting notes, a magnifying glass shows where in the transcript or your own raw notes it came from; for the transcript, a Transcript Summary popover sums up that part of the call and quotes the passage. Granola Chat answers carry inline citations to the meeting notes they draw on. The same move outside search, with a meeting as the source. The trail ends at transcript text: Granola saves no audio to play back.
Visit Granola ↗
Research Frame
Why this pattern exists, and where it breaks.
Map every claim to its contributing source spans; reveal the link on hover.
Chat citations are flat numbers — readers cannot verify which sentence is grounded where.
False precision: a visible source link can imply rigor that the underlying retrieval lacks.
Common questions
What people ask about this pattern.
When should I use the Citation Trail pattern?
+
Readers must verify which source supports which specific claim, not just the document overall.
When should I avoid the Citation Trail pattern?
+
The output is opinion or synthesis where granular source mapping would be misleading.
What problem does Citation Trail solve?
+
Sourced AI hides which fact came from which source behind aggregate footnotes.
Why is chat the wrong fit for this?
+
Chat citations are flat numbers — readers cannot verify which sentence is grounded where.
Grounded in Research
Not just opinion.
The papers, guidelines, and usability studies this pattern rests on.
Google PAIR team
Google Research · 2021 · refreshed 2026-07
Six-chapter first-party guide. Its Explainability + Trust and Feedback + Control chapters map almost one-to-one onto our Trust & Verification and control patterns; its confidence-display guidance ("determine if, when, and how to show confidence") is the softer default our confidence-signals avoidWhen encodes.
Raluca Budiu, Feifei Liu, Emma Cionca, Amy Zhang
Nielsen Norman Group · 2023
Diary study of real chatbot use: only ~4% of turns include a clarifying question and only ~22% of AI answers get fact-checked — the empirical backbone for asking-before-guessing and for making verification cheap enough to actually happen.
Caleb Chan
Nielsen Norman Group · 2025
Recommends against exposing step-by-step reasoning as a trust mechanism: displayed traces are typically post-hoc rationalizations, not the model's faithful process — lean on sources and stated limitations instead. Directly sharpened reasoning-trace's risks framing.
Related Patterns
Curated connections, not just neighbors.
Hand-picked relationships — what this pattern pairs with in practice, what it replaces, what it builds on, and what builds on it.
Often paired with
Confidence Signals
Confidence Signals add how sure each claim is, with a color-coded badge and a source count.
Often paired with
Reasoning Trace
Citation Trail shows which source backs each claim; Reasoning Trace adds the assumptions and inferences behind it.
Often paired with
Smart Document
Citation Trail ties each sentence to its source span; Smart Document recomputes the derived text when that source is edited.