Anthropic has expanded Claude’s capabilities to include custom charts, diagrams, and interactive visuals generated directly inside chat responses. The feature allows the AI assistant to produce tailored visual content on the fly, from flowcharts to data-driven graphs, without requiring users to leave the conversation window. For professionals and casual users alike, this marks a shift in how AI chatbots communicate complex information, although the feature comes with real constraints that limit its reach.
How Inline Visuals Work in Claude
When a user asks Claude a question that lends itself to visual explanation, the AI can now render a diagram, chart, or interactive graphic directly within the chat. These are not static images pulled from a database. They are custom visuals generated for the specific query, built in real time and displayed inline alongside text-based answers. A user asking about organizational structure, for instance, could receive a flowchart. Someone analyzing quarterly sales data might get an interactive bar chart.
The interactivity element is significant. Users can hover over data points, zoom into sections, or otherwise engage with the rendered output rather than simply viewing a flat image. If the first version of a visual does not quite match what the user needs, they can request changes through follow-up messages, and Claude will update the graphic accordingly. This iterative loop turns what might have been a one-shot illustration into a back-and-forth design process within the same conversation thread.
Once a visual meets the user’s needs, there are several export options. The output can be downloaded as an .svg file for use in presentations or documents, or as an .html file that preserves interactivity for embedding in web pages. Users can also save the visual as an artifact within Claude’s system for later reference. That flexibility matters for anyone who needs to move AI-generated content into a professional workflow, whether for a client report or an internal briefing.
Distinguishing Custom Visuals From Data Widgets
Anthropic draws a clear line between two categories of visual output in Claude. One category involves real-world data widgets, such as weather forecasts or recipe step-by-step guides, which pull structured information and present it in a formatted card. The other category, and the focus of this update, involves custom-built visuals like diagrams and interactive charts that Claude constructs from scratch based on the user’s specific question.
The distinction matters because it clarifies what Claude is actually doing when it produces a visual. A weather widget retrieves and formats existing data. A custom diagram, by contrast, requires the AI to interpret a prompt, decide on an appropriate visual format, and generate the underlying code to render it. That second task is far more demanding and far more useful for knowledge workers who need bespoke illustrations rather than templated data cards.
This two-track system also helps set user expectations. Not every visual response from Claude involves the same level of generation. When a user sees a recipe card, they are looking at a structured widget. When they see a system architecture diagram or a comparative bar chart, they are looking at something Claude built specifically for that conversation. Understanding the difference helps users know when they can push for revisions and when they are viewing a more fixed format.
Artifacts and the Question of Persistence
Custom visuals in chat sit alongside a broader feature set that Anthropic has been building out. The company’s Artifacts system, which is now generally available, allows Claude to produce persistent, shareable outputs including flowcharts, SVG graphics, websites, and interactive dashboards. Artifacts are designed for projects that need to live beyond a single conversation (offering a more durable home for complex work products).
The relationship between inline visuals and Artifacts raises a practical question for users: when should they use one versus the other? Inline visuals are fast and conversational, ideal for quick explorations or brainstorming sessions where the goal is understanding rather than deliverables. Artifacts, on the other hand, are built for outputs that need to be edited, shared with collaborators, or revisited days later. The fact that inline visuals can be saved as artifacts provides a bridge between the two modes, but the default experience is ephemeral.
That design choice carries a tradeoff. For individual users who want a quick chart to understand a concept, inline generation is efficient and low-friction. For teams working on shared projects, the ephemeral nature of chat-based visuals could create friction. A consultant who generates a useful process diagram during a conversation would need to actively save it as an artifact or export it before it becomes buried in chat history. The extra step is small, but in fast-moving professional settings, small steps often become missed steps.
Platform Limits and What They Signal
The most notable constraint on custom visuals is platform availability. The feature works only on web and desktop versions of Claude, not on mobile. For a workforce that increasingly relies on phones and tablets for quick reference and on-the-go decision making, this is a meaningful gap. A manager reviewing data on a commute or a field engineer checking a system diagram from a job site would not have access to these visual capabilities.
The web-and-desktop restriction likely reflects technical realities. Rendering interactive SVG-based or HTML-based visuals requires screen real estate and processing power that mobile interfaces handle less gracefully. But technical explanations do not change the user experience. Until mobile support arrives, the feature serves a subset of Claude’s user base, and that subset skews toward desk-bound knowledge workers rather than the broader population of mobile-first users.
This limitation also hints at where Anthropic sees the highest-value use cases for visual generation. Desktop and web users tend to engage in longer, more complex sessions. They are more likely to be analyzing data, planning projects, or building presentations. By launching visuals on these platforms first, Anthropic is targeting the users most likely to push the feature hard and provide feedback that shapes future iterations. It is a deliberate sequencing choice, not an oversight, but it still leaves a gap in the product’s reach.
What This Means for AI-Assisted Work
The addition of inline visual generation changes the nature of what users can expect from a text-based AI assistant. Until recently, chatbots were fundamentally text-in, text-out tools. Users could ask a question, receive a paragraph or two of explanation, and then either accept the answer or ask for a rewrite. Any diagrams or charts typically had to be created manually in separate software, even if the AI helped describe what they should look like.
With Claude now capable of rendering visuals directly in the conversation, that division between ideation and production begins to blur. A product manager can sketch out a user journey in prose and immediately see it translated into a process diagram. A data analyst can paste in a small table of numbers and get a quick visualization to sanity-check trends. The assistant becomes less of a passive explainer and more of an active collaborator that can propose, revise, and finalize visual artifacts in real time.
This shift could change how teams structure their workflows. Instead of waiting until the end of a research or planning process to commission charts and diagrams, users can generate them continuously as they think through a problem. Early-stage visuals may be rough, but they can surface misunderstandings or missing data before decisions are locked in. Over time, as users grow more comfortable iterating on visuals through conversation, the boundary between “draft” and “final” output may become more fluid.
At the same time, the feature’s constraints will shape adoption. The lack of mobile access means that many spontaneous use cases (sketching an architecture diagram during an in-person meeting, for example) may still fall back to whiteboards or separate tools. And because inline visuals are tied to specific chats unless saved or exported, organizations that want a canonical library of diagrams will need to build explicit habits around converting one-off visuals into longer-lived artifacts.
For Anthropic, custom visuals and the broader Artifacts ecosystem point toward a vision of AI that is less about answering questions and more about co-creating work products. If that vision holds, future updates are likely to focus on smoothing the handoff between quick inline sketches and persistent, shareable assets, as well as extending support beyond the desktop. For users, the immediate takeaway is simpler: the next time a concept feels hard to grasp in words alone, Claude can now draw the picture instead of just describing it.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
