> ## Documentation Index > Fetch the complete documentation index at: https://invoca-5bd45748-mintlify-17ed87db.mintlify.site/llms.txt > Use this file to discover all available pages before exploring further. # Identifiers: Avatar > Whether an agentic feature gets a visual mark distinct from a user's own avatar, and what that mark must and must not do. **No visual identity has been decided for any Invoca agentic feature yet.** This page proposes a framework for making that decision — it is not the decision itself. See [Coverage, stated honestly](/invoca-design-system/ai-experience/overview#coverage-stated-honestly). ## What it governs Whether an AI feature gets a visual mark — an icon, a monogram, an abstract shape — that stands in for it the way [Avatar](/invoca-design-system/components/data-display/avatar) stands in for a person, and what that mark may and may not imply. It governs the *decision to have a mark at all* and the *rules a mark must follow once one exists*. It does not govern the `Avatar` component itself, which already exists, ships unthemed, and is documented at [Components: Avatar](/invoca-design-system/components/data-display/avatar). If an AI mark is eventually built, it most likely reuses that component's rendering shape — a circular slot, a size token, a fallback order — but whether the content in that slot represents a person or a system is exactly the question this page answers. ## Vocabulary | Term | Means | Not to confuse with | | ----------------- | --------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **AI mark** | A visual element that identifies output as coming from an agentic feature, not a person | A **user avatar**, which identifies a specific human | | **Identity mark** | A mark that answers "who or what is this" | A **state indicator**, which answers "what is it doing right now" — streaming, working, done. That belongs to [Governors](/invoca-design-system/ai-experience/governors/stream-of-thought), not to this page. | | **Fallback** | What renders when no mark has been decided | Not the same as the `Avatar` component's own image → children → default-icon fallback order, which is a rendering mechanism, not an identity decision | ## Choosing a value | If you need to… | Use | Not this, because | | ----------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------ | | Show that a message or card came from an AI feature, not a person | A mark visually distinct from `Avatar`'s photo/initials pattern — for example a fixed icon glyph, never a photo slot | Reusing `Avatar`'s circular photo-or-initials pattern implies a specific person exists behind the mark | | Show the AI is actively generating or "thinking" | An outcome-state treatment on the relevant pattern page — see [Governors](/invoca-design-system/ai-experience/governors/stream-of-thought) | Animating or restyling the identity mark itself to mean "working." That collapses two meanings — who, and what it's doing — into one glyph | | Make several agentic features (Signal AI, a future assistant) feel like one product | Decide first whether Identifiers apply per-feature or once for all of Invoca's agentic surfaces — this is an open gap, not a default | Assuming a shared mark without deciding this question — see [Gaps](#gaps) | | Render an AI mark before any has been decided | No mark at all — a plain text label naming the feature | The default Person icon `Avatar` falls back to, which reads as "an unspecified human," the opposite of what an AI mark needs to say | ## Reference Nothing is decided or shipped. No Invoca agentic feature has a stated avatar, icon, or mark today — including Signal AI. This section will be generated from a real decision once one exists, the way [Color's token reference](/invoca-design-system/foundations/color#token-reference) is generated from source today. ## Relationships * **[Components: Avatar](/invoca-design-system/components/data-display/avatar)** — if an AI mark is built, it most likely sits in the same slot and inherits the same size and shape conventions. It does not inherit the photo/initials content model, which assumes a person. * **[Foundations: Color](/invoca-design-system/foundations/color)** — any color used on an AI mark is a semantic or component token, never a new hue. See [Identifiers: Color](/invoca-design-system/ai-experience/identifiers/color). * **[Foundations: Iconography](/invoca-design-system/foundations/iconography)** — if the mark is a glyph rather than an image, it comes from the closed 154-icon set or a formal proposal to it, per `TITAN-ICO-01`. See [Identifiers: Iconography](/invoca-design-system/ai-experience/identifiers/iconography). ## Accessibility * An icon-only mark needs an accessible name stating what it is — "AI-generated summary," not a description of the glyph's shape. * A mark is never the only signal that content is AI-generated. Pair it with text — the feature's stated name, see [Identifiers: Name](/invoca-design-system/ai-experience/identifiers/name) — so the disclosure survives for anyone who cannot see or distinguish the mark. * The mark itself must not be mistaken for a live status. If it moves or changes to signal "working," that is a [Governors](/invoca-design-system/ai-experience/governors/stream-of-thought) concern with its own accessible-name and live-region requirements, not this page's. ## Constraints | ID | Constraint | Rationale | | ------------------ | ------------------------------------------------------------------------------------------------------------------------------------ | ---------------------------------------------------------------------------------------------------------------------- | | **TITAN-AIAVT-01** | An AI mark, if built, renders through Titan's existing avatar and icon token surface — never a new hardcoded image asset or literal. | A one-off asset is invisible to the token audit and forks the moment a second AI mark is needed. | | **TITAN-AIAVT-02** | An icon-only AI mark carries an accessible name identifying the content as AI-generated, not the feature name alone. | A screen reader user needs to learn the content's origin, not just its label — the two are different facts. | | **TITAN-AIAVT-03** | An AI mark is never the sole signal that content is AI-generated. Pair it with text or the feature's stated name. | Matches the overview's disclosure requirement — a shape a user has to learn is not disclosure at the point it matters. | | **TITAN-AIAVT-04** | Where no mark has been decided for a feature, its output renders with no avatar at all, never a placeholder or a reused default. | A placeholder reads as a decision to the next person who sees it. An absence reads as undecided, which is accurate. | | **TITAN-AIAVT-05** | Once decided, a feature's mark is fixed across the surfaces that feature appears in — it is not restyled per screen. | Recognition is the entire value of a mark. A mark that changes shape per surface teaches the user nothing. | ## Divergences Not applicable. Nothing is shipped yet to diverge from. ## Gaps * Whether Identifiers apply per-feature or once for all of Invoca's agentic surfaces is undecided — flagged on the [overview](/invoca-design-system/ai-experience/overview#gaps) and unresolved here too. * Whether Signal AI, which runs as a background analysis job rather than a chat surface, needs a visual mark at all is undecided. * What shape a mark should take — icon, monogram, or an abstract form — is undecided. ## Related [Identifiers: Color](/invoca-design-system/ai-experience/identifiers/color) · [Identifiers: Iconography](/invoca-design-system/ai-experience/identifiers/iconography) · [Identifiers: Name](/invoca-design-system/ai-experience/identifiers/name) · [Identifiers: Personality](/invoca-design-system/ai-experience/identifiers/personality) · [AI Experience overview](/invoca-design-system/ai-experience/overview) · [Components: Avatar](/invoca-design-system/components/data-display/avatar) · [Foundations: Color](/invoca-design-system/foundations/color) · [Foundations: Iconography](/invoca-design-system/foundations/iconography) ## Why it works this way An avatar answers "who is this" the instant it renders, before any label is read — which is exactly why it is dangerous applied carelessly to AI output. The `Avatar` component's own content model (a photo, or initials drawn from a name) is built entirely around representing a specific person, and reusing it unmodified for a model's output borrows that implication without earning it. Treating the mark as an identity question, separate from state and separate from color, is what keeps this page from re-deciding Governors' or Trust builders' job. A mark that also tries to say "working" or "uncertain" is carrying meaning it was never designed to hold, and the first time it needs to say two things at once, it will say neither clearly.