@ForwardEditor I want to try it out on the peloton bike later. Probably a good method to use it since you can have a tray that holds your phone in case you need to handle or preview anything. Would be more difficult when I’m mountain biking hah.
What about your session agent as the orchestrator and you say '/goal orchestrate a workflow of an agent department for this repo with one Opus 4.8 team lead agent that evaluates open GitHub issues. Take inventory of issues and identify dependencies on other issues and track in todo.md and assigns a team of 1 execution agent with Sonnet 5 and 1 code review agent with Sonnet 5 per GitHub Issue. Have the execution agent run /implement on his assigned issue and notify you on code complete to trigger the review agent to run /code-review on the same issue while the execution agent has a 5 minute loop pollling for code-review completion by the review agent and fix findings and report back issue is complete to you and modify todo.md , exit the goal when all issues on the todo are complete or after [x] hours”
Used wisper flow and it sounds too verbose, probably stupid. Just my only thought to flow AFK through a large group of issues while keeping main agent context clean with all the code execution and code review happening by subagents in their own context. I’ll wait to see a better reply because mine is surely terrible hah.
@jxnlco@ajambrosino Great video, referenced it in my lunch and learn for today going over new features with product and devs. Hope to see more features marketed with you.
ChatGPT Voice is now in the desktop app.
Control your computer and direct multiple agents running in ChatGPT Work or Codex, using just your voice.
It's powered by GPT-Live, so it can speak, listen, and coordinate work in the app at the same time.
Rolling out globally today on macOS and Windows to Plus, Pro, Business, Edu, and Enterprise plans.
@Rasmic First, pumped to see videos on your local setup. Second, did you choose Windows instead of Ubuntu so you have a GUI and use tools like Computer Use?
Too much curated content out there. Even if it’s genuine and high quality it starts to insensitize you from that raw, unedited humanity, like a live mentorship. And that’s not a knock to your content I love your content, it’s the general AI influencer content as a whole. It makes me miss when people just went live on Twitch. Got to see a human unapologetically being themselves, delivering the content you want.
100% ageee with you. It’s wild how everyone sees the hype of new OW models (or even competing US frontier models) and tell everyone to jeet to the new thing like it’s not gonna be another new model in 3 weeks.
These OW models just push OpenAI, Anthropic, and now SpaceXAI to keep pushing because challengers are on their heels.
It’s also not like a $100/$200 subscription (that is subsidized heavily) is killing your budget if you’re using it right to give you back time and the ability to ship.
Get a $20/$40 Kimi sub, build some prototypes then go back to your tried and true stack and challenge them because it’s guaranteed their next model can out perform the current hype model.
Beginning July 20, Claude Fable 5 will be included in all Max and Team Premium plans, at 50% of limits.
Pro and Team Standard users will continue to have access to Fable via usage credits, and will receive a one-time $100 credit.
Demand for Fable has been challenging to predict, which is why we rolled it out to subscription plans in stages, extending access several times as we secured additional capacity.