Twitxy Saved

Matt Green

Verified account Speak

@IAmMattGreen

raging against entropy. AI Platform @speak
California Joined September 2009
1.2K Following 1.1K Followers 3.2K Posts

Latest posts 5

Matt Green Verified account @IAmMattGreen
been using dot since launch and used it this way at first but quickly found the "single thread for everything" cumbersome and unwieldy. Multiple threads are really helpful for keeping thoughts separate and effectively parallel tracking tasks
Peter Gostev Verified account @petergostev 15h
If you are first getting access to a Dot, the way I think about it is that it is a perfect ChatGPT killer - I don't think I've opened the old-school ChatGPT once after I got Dots.

The key mindset shift is that you need to stop worrying about different threads, you can talk to it on any topic one message after another and it won't lose track. I had conversations going about buying insurance, looking for SSD drives, questions about Anthropic IPO and god knows what all in one stream and it never lost track.

Then on top of it, it has its own computer and can actually do things. There are still many details to be ironed out, but it can remind you, do things on your behalf (form filling is my favourite), and set up codex threads if you need to do something more complex.

It can also be nicely proactive - it has called out that I needed to fill in an arrival form for my travel that I had no idea about, or deadlines coming up from the emails that I've missed, or that my location in the calendar didn't match what we have agreed via LinkedIn.

It is still not perfect - integrations and handoffs are still somewhat painful, permissions can be annoying and proactivity could be better - but as a ChatGPT killer it really feels like the final form of what an assistant should be like.
View quoted post on X
Matt Green Verified account @IAmMattGreen
Fast execution is no longer a competitive advantage, it is a requirement. Questions to ask:
- how fast are they shipping?
- are non-technical people shipping code?
- How automated is their deployment and recovery flows?
- are PR's getting opened and merged at night?
what else?
Matt Green Verified account @IAmMattGreen
This will be huge for codebase gardening tasks like security scanning, docs cleanups, and more!
OpenAI Developers Verified account @OpenAIDevs 2d
Not every workflow requires an immediate response.

For single- or multi-turn workflows that can wait, save up to 40% per token than Standard on GPT-6.2 Astra or GPT-6.2 Sol.

In Responses API background mode, choose a 2h or 8h target for the first response. These discounted windows are now in public beta.
View quoted post on X