johaugum

Joined 8 years ago

Karma

Submissions

Member since

Jul 17

Recent submissions

Kind of, Swappa has new sealed units with the original warranty (and the warranty timer only starts when you set up the device), but indeed these are resale and not directly from Apple.

Comment · 2 months ago

Are you on the recent update that handles long chat sessions much better? That made a big difference. Biasing towards lightweight and fast tool alternatives (uv over pip, bun over npm, ripgrep over grep, etc).

If that still doesn't resolve it, a free tier VM might be the move. I suspect OpenAI will be bundling some kind of VM with their new Astra release, which Sama said yesterday will be released soon (this week maybe?). If this new model supports much longer running tasks, you'll want a dedicated box that can run 24/7 even when your Neo is closed and asleep anyway.

Comment · 2 months ago

Mac Neo can be had for $600 now, and it runs Codex incredibly well!

But even on the Windows machine, if you Tailscale into a free tier VM for the workloads, you'd probably have a much better time.

Just leave Codex itself on your machine, if you set up computer use it will happily set up all of this for you while you have lunch.

Comment · 4 months ago

Siri was invented by a Norwegian (Dag Kittlaus), named after his own daughter!

Steve Jobs later called and bought his company.

Comment · 5 months ago

Source?

Comment · 7 months ago

On index benchmark results:

>Still not quite as fast as Go, but it's close. Honestly, I don't know why the memchr-based implementation is still slower than Go's assembly here, but I decided not to pursue it any further.

Libc call overhead. Your version is using an already-fast memchr, but it still has to cross a general libc boundary and then do the pointer/index around it.

Index with a 1-byte needle collapses to IndexByte, and on arm64 that goes into Go asm.

Comment · 8 months ago

>We already see some of the most popular coding agents today – like Claude Code and OpenCode – send these accept headers with their requests for content.

Expect this to be used to block agent traffic

Comment · 9 months ago

As you mentioned, that depends on what you mean by planners.

An LLM will implicitly decompose a prompt into tasks and then sequentially execute them, calling the appropriate tools. The architecture diagram helpfully visualizes this [0]

Here though, planners means autonomous planners that exist as higher level infrastructure, that does external task decomposition, persistent state, tool scheduling, error recovery/replanning, and branching/search. Think a task like “Prompt: “Scan repo for auth bugs, run tests, open PR with fixes, notify Slack.” that just runs continuously 24/7, that would be beyond what nanobot could do. However, something like “find all the receipts in my emails for this year, then zip and email them to my accountant for my tax return” is something nanobot would do.

[0] https://github.com/HKUDS/nanobot/blob/main/nanobot_arch.png

Comment · 9 months ago

> a lot of people seem to have found a really nice alternative to actually having to use their brains during the day.

Or have they have found a way to use their brains on what they deem as more useful, and less on what is rote?

Comment · 9 months ago

Skimmed the repo, this is basically the irreducible core of an agent: small loop, provider abstraction, tool dispatch, and chat gateways . The LOC reduction (99%, from 400k to 4k) mostly comes from leaving out RAG pipelines, planners, multi-agent orchestration, UIs, and production ops.

Comment · 11 months ago

Can confirm, has been flawless for me. I waited until 2 weeks after release to upgrade, possible I avoided some initial friction that way.

The only device I’ve found more sluggish after this recent OS upgrade is my Apple Watch Ultra (gen 1).

Animations when navigating the OS are noticeably sluggish where the previous version was smooth as butter. This degradation has persisted through multiple minor version updates since, so it seems to be permanent.