Search on Codex is not good, you hardly get the accurate results. One important issue is it's not easy to initiate new conversations in codex on mobile.
This leads to starting a conversation in chat/work and while working in codex if you search for something you can't find it and later you have to create a new conversation asking to find the missing conversation and then find that it's on work.
One good data to pull is the number of conversations the users are having to find the previous conversations, this will make the user experience better.
Now in beta: Businesses in ChatGPT.
Claim your business, manage your information, and get insights on how people engage with you.
Eligible businesses can also build a Business Agent to answer customers’ questions and help them explore what you offer.
https://t.co/N2GEZX0mFT
Now in beta: Businesses in ChatGPT.
Claim your business, manage your information, and get insights on how people engage with you.
Eligible businesses can also build a Business Agent to answer customers’ questions and help them explore what you offer.
https://t.co/N2GEZX0mFT
@JustinBleuel@ChatGPT The voice chat is always glitchy and definitely not a smooth experience while driving and it’s always a question if the voice is connected or not. Switching between codex and chat is too confusing.
@StefanoGPT I am running a very taxing job just to help you out here, interestingly I feel this version of the app is very efficient as it hardly consumes any tokens and I’m worried if it’s actually getting the job done, I’m on - 26.901.51231
1. Make comuter use / browser use stable for long running tasks without any disconnects
2. If we change the model preference for a running task, the same gets applied for all new, rather there should be a default configuration that applies for every new conversation and not the last one.
Giving a rate stop for the token usage.
I was working on something which needs computer usage across multiple tabs and decision making across all of these different tabs, when I was done with one task I asked codex if there is an efficient way to do it it came back with bunch of suggestions which were quite good. Imagine I give a rate limit that you need to achieve 100 steps of this within 1% of usage codex should figure out the efficient way and this can be a default too and today I’m achieving this via prompting but it takes lots of efforts
Secondly I want to run these in loops and go to bed but making sure it just doesn’t exhaust all my rate limits by time I wake up, so an option to rate limit the token usage will be quite helpful
Most founders copy/paste their privacy policy from Stripe or ask Claude to draft an MSA at 2:00 AM, then pray they don't get sued.
You can stop doing this.
We built a free, CC0-licensed legal template library for startups:
-MSAs
-Employment agreements
-Privacy policies
-More...
And because many of you on here asked, here is an agent-friendly link to where these are posted on GitHub:
https://t.co/GF62d39zHo
Most founders copy/paste their privacy policy from Stripe or ask Claude to draft an MSA at 2:00 AM, then pray they don't get sued.
You can stop doing this.
We built a free, CC0-licensed legal template library for startups:
-MSAs
-Employment agreements
-Privacy policies
-More...
And because many of you on here asked, here is an agent-friendly link to where these are posted on GitHub:
https://t.co/GF62d39zHo
I used codex to solve multiple issues on my mac like my mic got stuck and then internet issues and more sorts, I one time used Hermes to solve when codex was not responding too. And I do use codex to enable and disable certain features on mac too. If we expand this and can do the same across mobiles and other devices we do tech support will be gone soon