baalimago

19 ms·

1.
Scanning Sweden on a Raspberry Pi(lorentz.app)
2 points by baalimago 4d ago|0 comments
2.
by baalimago 6d ago
I found that trying to keep up with their API updates and changes is more trouble than its worth, even with AI. The agents would need to reverse engineer from the OpenAI SDK source anyways, so why not cut the middleman? If I wanted somethin
3.
by baalimago 6d ago
I've had great success with the OpenAI agents SDK [0]. This way I've been able to build the sandbox + slack + knowledge-bank integrations independently and be very strict with what I expose to OpenAI. Looking at the Agents API, it
4.
by baalimago 7d ago
So it's without collisions, I take it? I got confused by the premise. To the fully blind person, the speed of light does not matter. He'll hit the object 5 meters away from him regardless if he sees it or not (as will everyone els
5.
by baalimago 7d ago
I suspect I would get greater value by simply duct-taping two iPhones together into a fold.
6.
GitHub CI outages in EU due to status 401 on public repos(github.com)
3 points by baalimago 8d ago|0 comments
7.
by baalimago 11d ago
A bit narrow definition of a "cloud", no..? More like a microservice server on rails Bits missing from cloud being global CDN (selfhosted), routing, regional sharding, etc.
8.
by baalimago 13d ago
When writing agentic apps (apps which embed an agent for semantic work), exposing the same tools which are injected into the agents as an MCP allows for practically the same functionality to be debugged and run in a normal agentic harness.
9.
Show HN: Skillbased Versioning v1.0.2(skillbased.space)
1 points by baalimago 18d ago|0 comments
10.
by baalimago 21d ago
They ensure the company is funded.
11.
by baalimago 21d ago
Good luck finding investments by going to events in socal
12.
by baalimago 21d ago
Programming, much like artisinal chairbuilding, will be demoted to a hobby or niche. Software engineering will remain.
13.
by baalimago 23d ago
I am working on a very similar mcp server which allows agents to communicate and share notes without race conditions, using git and s3 as the driver [0]. Funny how "inventions" like these pop up independently. [0]: slivingdoc.dev
14.
Show HN: Slivingdoc, conflict-resolving notebook for agents w. S3 backend(slivingdoc.dev)
1 points by baalimago 1mo ago|0 comments
15.
Bubblewrap is the only sandbox you need(lorentz.app)
2 points by baalimago 1mo ago|0 comments
16.
by baalimago 1mo ago
I've found that keeping very healthy test coverage solves issues like this. LLMs are very good at validating their own implementations with TDD.
17.
by baalimago 1mo ago
Boring is better. Perfection is the enemy of good.
18.
by baalimago 1mo ago
So without sandbox: it doesn't apply. Without explicit tool approval step: it doesn't apply. Response in xml: it doesn't apply. What is the point?
19.
by baalimago 1mo ago
Next month perpetually
20.
by baalimago 1mo ago
Give it an AI-bubble pop and these will be flooding the market.
21.
by baalimago 1mo ago
"Prevent sub-optimal code by manually retyping compiler-generated assembly" I don't think this is a practice which will be sustainable for very long
22.
by baalimago 1mo ago
I've gone another route when I found myself simply replying with an investigation made from an AI: Actively encouraged my coworkers to ask the knowledge-bank AI I've setup instead of asking me. It has read-only credentials for all
23.
by baalimago 2mo ago
This: "(b Box[T]) Map[U any](f func(T) U) Box[U]" is the type of cognitive weight I was happy that Go avoided.
24.
by baalimago 2mo ago
Personally I'm hoping for EU to step up a bit
25.
by baalimago 2mo ago
New Deepseek models are like Christmas for me. Really big fan of low cost API models, noone does it better than DS. Until VRAM price is low enough to run models locally, this is the way to go. The subsidized subscription model won't la
26.
by baalimago 2mo ago
Looks like they will release an updated version of deepseek-v4-pro soon, which most likely will beat kimi k3 at both intelligence and cost (judging by the vast improvements to dsv4 flash)
27.
by baalimago 2mo ago
Very promising. So it will both keep the speed and reduced price, yet exceed performance of the quite sufficient deepseek-v4-pro? Should be extending the lead in intelligence/cost index, as deepseek-v4-flash already were the most price
28.
by baalimago 2mo ago
It seems like Europe, yet again, have to stand up for ethics, morality and try to deflect corruption. It's so tiresome.
29.
by baalimago 2mo ago
No we kept prompts as is, just swapped model. The prompt is already quite optimized for the task. How would updating it possibly make a more intelligent model spend less tokens than a less intelligent model? Care to elaborate?
30.
by baalimago 2mo ago
We swapped an internal system from gpt-5-mini to gpt-5.6-luna and saw no benefit but 4x cost. Sufficed to say: we swapped back to gpt-5-mini.

More ›