20 posts tagged with "ai"

We've updated the Carbon Voice MCP server with a big batch of new capabilities, which means you can use Claude and ChatGPT to do more with Carbon Voice than ever before.

If you've already connected Carbon Voice to Claude or ChatGPT, the new capabilities are there automatically, with nothing to install or update. If you haven't connected yet, follow the setup instructions for Claude or ChatGPT, and you'll have all of it from the start.

By default, a text message you send plays back to the other person in a standard text-to-speech voice. Clone your voice and it sounds like you spoke it instead.

You can now connect ChatGPT to Carbon Voice. Capture a voice memo on the go, then open ChatGPT and ask it to "turn my last voice memo into a blog post." Or turn a whole conversation into structured documentation: "Turn our conversation 'Acme Investment' into an investment memo."

Carbon Voice 3.0 is our biggest update yet, available now in preview.

Every conversation holds context, decisions, and next steps — and too often they're lost. 3.0 turns talk into insight, action, and knowledge, so a conversation becomes living documentation and clear next steps instead of ending. Three new tools make that work: AI Chat, Action Items, and Notebook.

You can now write your own AI prompts to turn conversations and voice memos into whatever structured output you need.

If you use Carbon Voice alongside an AI assistant like Claude or Cursor, you can now ask those tools to analyze and synthesize your conversations and voice memos directly. No copy-pasting, no exporting — just context.

A fast new way to catch up on a conversation, built straight from your feedback: turn every message in a conversation into bullet points at once.

At least it feels that way. We rebuilt the conversation experience around your feedback, and the result is a subtle change that makes conversations feel fresh and much easier to follow.

You told us that customizing your text-to-speech voice matters, so you can now set a preference for the voice used when text is read aloud.

People are getting a lot out of AI Actions, so here are the seven most popular ones based on real usage data — what they're good for, and what each actually does.

Global Voice is a new way to talk to someone who speaks a different language. Your messages are translated automatically when the recipient doesn't understand your language — they can read the translation, or hear you speak their language in your own cloned voice. It also turns your text messages into voice messages, played back in that same cloned voice.

Whether it started as a team discussion, a voice memo on the go, or a brainstorm, the Google Slides add-on takes you from idea to a draft presentation.

This release is about catching up quickly and turning discussions into structured updates.

Let Carbon Voice write your first draft. AI Actions let you speak an idea and get structured text back, from a voice memo or from any message in a conversation. There are hundreds of possibilities — here are nine our own team keeps coming back to.

AI Actions turn any voice memo or message into something you can use: an email, a to-do list, a blog post, a sales summary, a bug report. Capture with voice, structure with AI.

Another round of features for capturing, organizing, and sharing your voice messages.

Want your AI summaries alongside the original message? Prefer to hide audio transcripts so more fits on screen? New preferences let you see a conversation the way that works best for you.

In response to your feedback, AI summaries are now on by default across all conversations. Any message that has a summary — typically messages longer than 100 characters, on accounts with an AI plan — will be summarized, so you can grasp the essence of a long message without reading the whole thing first.

You can now see what a message is about without opening the app — so you get the gist right away, and stop guessing which notification to check first.

Your feedback inspired some big changes. This release adds the things you've come to expect from any communication tool, plus a few that only make sense in a voice-first one.