"Walk me through it" - Live screen-sharing + voice guidance to help navigating complex workflows
What variant of Codex are you using?
App
What feature would you like to see?
I’d like Codex to support live screen-sharing with voice guidance.
Use case: while I’m working in a complex UI — for example AWS, system settings, a web app admin panel, or an online form — I want to stream my screen to Codex and have Codex talk me through what to do step by step.
Today, I have to send screenshots manually. Codex then says things like “click here,” “go back,” “open this menu,” or “type this value.” That works, but it is slow and error-prone. If I click the wrong thing, I need to send another screenshot and wait for the next instruction.
The ideal experience would be closer to Apple Support screen-sharing: Codex can see what I’m seeing in real time, and we can speak vocally while it guides me through the task.
Example:
I recently needed to configure an AWS function to allow email sending. The AWS interface is complex, and I had to keep sending screenshots to Codex. Codex was able to guide me successfully, but a live visual stream plus voice would have made the workflow much faster.
Other examples:
- Filling out online documents or forms
- Configuring SaaS/admin dashboards
- Debugging local app or system settings
- Walking through cloud infrastructure setup
- Helping non-technical users complete technical workflows
Feature request:
Add a mode where Codex can observe a live screen stream and provide real-time spoken guidance, without needing manual screenshots at every step.
Additional information
Additional Information
I think the impact of this feature extends well beyond software development.
Today, if someone gets stuck using a website or application, their options are usually to search the web, watch YouTube videos, ask a friend, or call customer support. A live screen-sharing mode would allow Codex to become the expert sitting beside the user, guiding them through the task in real time regardless of the application.
This would be valuable for technical workflows such as:
- AWS, Azure, or GCP configuration
- Kubernetes and DevOps tooling
- Local application setup and debugging
- System settings and operating system configuration
- Enterprise SaaS administration
But I think the larger opportunity is helping people navigate everyday digital tasks, for example:
- Filling out government forms
- Completing tax returns
- Applying for visas or passports
- Filing insurance claims
- Online banking
- Enrolling children in school
- Setting up a new computer or smartphone
- Recovering access to online accounts
- Configuring printers, Wi-Fi, or email
- Completing identity verification
- Scheduling medical appointments
- Filling out complex online forms
This could be especially transformative for users who are not technically confident. Rather than trying to explain what they see or send screenshots back and forth, they could simply share their screen and have Codex guide them conversationally, one step at a time.
I imagine this as an optional mode that the user explicitly enables. Codex would observe the user’s screen and provide real-time voice guidance while the user remains in control of the keyboard and mouse.