[FEATURE] Voice mode on claude.ai web: show diagrams and files Claude creates during the conversation

Preflight Checklist

• I have searched existing requests — related: #87294 (on-demand screen sharing in voice mode on iOS) and #98918 (two-way voice mode in Cowork). See Additional Context for why I'm filing this separately.
• This is a single feature request (not multiple features)

Problem Statement

I use voice mode on claude.ai web to learn technical topics. When a flow or architecture gets complicated, I ask Claude to draw a diagram — and in voice mode it can't reliably show one.

In one turn, Claude tried three times and every attempt failed: creating the SVG file failed, reading the PNG it rendered with code failed, and sharing the file failed with an error saying the tool isn't available during a voice call. Claude then told me "sending images is unstable" and fell back to describing the diagram out loud. Earlier in the same conversation, a diagram had been shown successfully, so the behavior is inconsistent and hard to predict.

Proposed Solution

Let diagrams and files that Claude creates during voice mode show up on screen in the same conversation:

• Render them in the conversation while I keep talking, or
• If that's not possible during the call, queue them and show them as soon as the voice session ends.
• If file output really isn't supported in voice mode, tell me once up front, instead of several failed tool calls.

Alternative Solutions

• Leave voice mode and ask again in text: works, but breaks the flow and I have to repeat the request.
• Ask for a text outline in voice and the diagram later: workable, but the diagram is exactly what I need at that moment.

Priority

Medium - Would be very helpful

Feature Category

Other

Use Case Example

1. In voice mode, I ask Claude to explain how short-term memory gets written into long-term memory in my project.
2. Claude says "let me draw it" and the diagram appears on screen while it keeps explaining by voice.
3. I point at a part of the diagram and ask a follow-up question, still in voice.

Additional Context

Environment: claude.ai in the browser on macOS, Sonnet 5.5, voice mode on.

I know this is about claude.ai rather than Claude Code, and #87294 was closed for that reason. I'm filing it because the failing steps are the same file-creation and file-sharing tools, and voice mode for Desktop/Cowork is already being discussed here (#98918). Happy to move it if there's a better place.

Thanks for considering it!