rainwater

Rainwater et al.

[Marq @marquisehurtt]

asked o8 where this week's work actually went and it drew the answer right in the workspace

[Marq @marquisehurtt]

https://t.co/ANtg33G3Vs | https://t.co/0ltAVYGZGU

[Marq @marquisehurtt]

Sign in to @o8dotrun with ChatGPT subscription. Will release in the next ship with some clean up fixes for space conservation I spoke about earlier. Bless you guys and thanks for keeping up with the build seriously. 🏄🏽‍♂️

[Marq @marquisehurtt]

adding opus 5.5 and 6 sol + imessage group support for symon, patch will release tonight.

Lets GOOOO!

[o8 @o8dotrun]

Waiting is optional. o8 0.1.760 adds a persistent control for pre-warming Brain runtimes, so you can turn early startup off.

With warmup off, o8 keeps that choice during pool refill. A Brain runtime then starts only when needed.

A mission effort level now stays fixed through launch. For managed Brain answers, Auto stops with an error if its route is unavailable. It does not switch to another route.

The public changelog now keeps vendor names. It also replaces a model name as a whole, so words do not run together.

Recorded claim and citation answers can replay with labels. Use --label all-recorded to include every recorded answer.

Startup, answer routing, and release notes now show the choice that actually happened.

[Marq @marquisehurtt]

For my Codex users:

o8 Mobile Voice Agent (Symon) can run on your ChatGPT subscription, not your weekly usage.

Jev is already being testing to make tool selection better for the voice agent during conversations.

Working working working.

[Marq @marquisehurtt]

Mobile coming at you fast! 💨 @o8dotrun

[o8 @o8dotrun]

One task, three workers, three worktrees. Opus 5 split it into packets, OpenCode 2 workers on DeepSeek 4.1 Flash built the pages in parallel, and I reviewed every diff before merging what passed. All of it on the record. Open source, https://t.co/DyTSI14ycg

[Marq @marquisehurtt]

Got the vids back working! LETS GOOO!

(I see the ui lol model text issues on expand, and the transcript formatting for workers. <3)

[Marq @marquisehurtt]

o8 is the substrate.

This is Astra checking the inbox the see Fable and other Astra backlog issue messages, lol simply put you don’t have to be inside of o8 for your agents and workflows to use it and coordinate.

Just point your agent at it, any issue tell it file it in the public repo, we will fix it in a day.

I’m on a Mac pass to make it’s load even smaller so running it is a breeze in background… pretty much forever.

[Marq @marquisehurtt]

YANMC.

YOU ARE NOT MY CUSTOMER.

One of the most important things you can learn as a founder (of anything) is that not everyone needs to understand, like, want, or buy what you’re building.

Stop burning time trying to convert people who were never your customer in the first place.

There are enough people in the world who have the exact problem you solve, and see value in the way you solve it.

Find them then build for them and always listen to them bro.

The wrong customer will make you question a great product.

Don’t fall into that trap.

[Marq @marquisehurtt]

Hermes Agent Desktop added profile-scoped skills, tools, and MCPs.

I would build those profiles around blast radius, because a research agent with browser access and an agent that can change production need different defaults.

[Marq @marquisehurtt]

https://t.co/UnjazBmv5W

[Marq @marquisehurtt]

I put tomorrow's first task where tired me can find it without making another decision

[Marq @marquisehurtt]

https://t.co/mMdY6ZJU0c