Talk/Design mode

Feature request for product/service

Chat

Describe the request

I must talk to (voice mode) the openAI chat bot about design ideas for about 2 hours a day, often when traveling and not able to use a computer.
I find i can talk to the agent about something i want to create and how i want a screen to work. Its a brain bump of what im trying to achieve. The AI agent asks me questions what i mean as it knows about the whole project. At the end of the chat (often +30 mins ) I have a plan of how to implement and i use this to feed into cursor or another coding agent.

I would be great if i could have think conversation and move this talk/design chat straight into a plan mode, then implementation.

Anyone else work like this?

Thanks Dan

hey @Dan_Cluderay, love your process, I think mine is kinda similar for some project, honestly, what I defaulted to was to use Cursor Cloud Agent directly from my phone to plan and prepare deeper session, I find it awesome, either i’d be in transport or out for groceries it’s working really fine for me! https://cursor.com/agent

I’ll try that @Tom_Coustols

And there’s a voice mode too!

Describe the request

I’m a product founder rather than a traditional developer, and I believe AI coding is evolving beyond chat windows.

I’d like Cursor to support an optional full-duplex conversational voice mode.

Rather than typing prompts and reading long responses, I want to work naturally by talking to my AI developer while watching the code and browser update in real time.

For example:

Me:

“Make the pricing cards feel more premium.”

Cursor:

“I’ve increased the corner radius, reduced the text density, and added a subtle shadow. The browser has refreshed—have a look.”

Me:

“The shadow is a little too strong.”

Cursor:

“No problem. I’ve reduced the opacity by 40%. Better?”

The agent should also speak meaningful progress while working:

  • “I’ve finished the Salesforce authentication.”
  • “I’m now implementing contact synchronisation.”
  • “I found an issue with the API response and I’m fixing it.”
  • “I’ve completed the changes. Would you like me to commit them?”

This shouldn’t be a screen reader or text-to-speech feature.

It should feel like collaborating with a senior developer sitting beside you.

Why this matters

When using AI coding agents I’m already watching:

  • the browser,
  • the running application,
  • the terminal,
  • the code editor.

Reading another large stream of text creates unnecessary cognitive load.

Voice allows me to keep my eyes on the product while naturally directing the agent.

I believe this is where AI-assisted software development is heading, especially for founders, designers and product builders who think conversationally rather than line-by-line in code.

I’m the same. Ive been a developer for +27 years and the rate of features I produce now is crazy.
Voice only has doubled my output and have a greater shared vision with the AI agent so it knows what i want

Honestly, I’m waiting for Grok voice mode to get added directly into Cursor so I can have side conversations about the current prompt or codebase. That way the voice chat still has full context, but it doesn’t burn through tokens the way a full voice back-and-forth inside Cursor would.