Couple thoughts on Astra (vs Fable):
I think for those that have been using all the recent models on coding related tasks, Astra does not feel like the large step that the release seemingly announces. It is very good tho in:
- Computer Use
- Writing and communicating clearly and concisely
- Understanding intent
I am sure for other use cases outside of coding this will make the capability jump be more tangible. For me it means I am still running Astra, Fable, Grok and Luna Max as daily drivers depending on use case or to check each others work.
It is the same AGI we have had for a while to me. Nothing changes other than progress progresses.
Not quite sure how the currently largest private company on earth buying another company supports open source.
But seeing continous insane competition is a good thing.
@mattpocockuk Luckily, this also means it is fine every now and then to not keep up with the frontier. For those that have been - happily - trying to do that over the past 36 months or so, its ok to take a break of the "Whats the latest on X?" treadmill.
Stripe CEO Patrick Collison and Mario Draghi have created the Rhine Group, an organisation for European revival.
They state that decline is not inevitable, and instead Europe should focus towards a common destiny where we have the capability to compete, build and grow.
@chamath It is organizations of every size doing the rollout. The distribution is not only skewed across use cases, it is probably even more skewed across internal usage of companies. For every tokenmaxxer there are 30 token catch up people.
@mattpocockuk’s skills have become my daily driving workflow. Especially wayfinder and prototype are extremely helpful for aligning intent and improving outcome quality. Definitely give it a try!
This is genuinely suprising as it is exactly our current numbers and trajectory across the company.
The challenge here will be to sustain and allow the top1% to continue on that usage and then how to incorporate that into revenue generation.
Will be an exciting rest of the year for sure.
Codex on windows just keeps having extremely long read/write/patch tool usage times. It is actively complaining about the environment taking long. WinDefender for turned off, OneDrive/GDrive off. What else is there to do? @OpenAIDevs@thsottiaux
@JayaGup10 Great take. What Fable is demonstrating is to reach for full autonomy at any cost. For steered use the cost/result outcome seems to be less worth it.
@OpenAIDevs How can we turn off the "out of credits" warning for every workspace member in our business account? It is extremely annoying to deal with.
- 4.7
- 5.5
- Cursor for 60b
- Cohere&AlephAlpha
- Workspace Agents
- Claude Design
- Salesforce Headless
- Flipbook streaming
Just one week, right? Just wondering when you talk to colleagues about the week before, does it feels like a long long time ago as well?
@OpenAIDevs I can connect to our tools via MCP in Codex, but I cannot use them in Workspace Agents. What could be the difference? Any docs on what to do to enable apps via mcp for workspace agents?