Hacker News: baalimago

New comment by baalimago in "CEO fired developers to make room for AI. Developers create open source AI CEO"

baalimago — Thu, 27 Aug 2026 04:53:56 +0000

They ensure the company is funded.

New comment by baalimago in "CEO fired developers to make room for AI. Developers create open source AI CEO"

baalimago — Thu, 27 Aug 2026 04:53:21 +0000

Good luck finding investments by going to events in socal

New comment by baalimago in "It's Not the End of Programming"

baalimago — Wed, 26 Aug 2026 11:56:33 +0000

Programming, much like artisinal chairbuilding, will be demoted to a hobby or niche.

Software engineering will remain.

New comment by baalimago in "Walgit – a Git server that is one binary in front of an object store"

baalimago — Tue, 25 Aug 2026 04:32:06 +0000

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

Show HN: Slivingdoc, conflict-resolving notebook for agents w. S3 backend

baalimago — Mon, 17 Aug 2026 13:21:33 +0000

Agents needs to store context, and share context. We haven't yet figured out how to do muli-agent swarms in a good way, but letting agents synchronize durably seems to be a key point, a problem which needs a solution regardless. So I decided to separate the concerns and tackle this bit before trying to build a swarm.

Slivingdoc is git + s3 combined into a "hot path". Conflict resolution via git, persistent storage + backup via s3. Agents + humans see gitlike-semantics, without needing to worry about the complexities. See website above for details.

Intended to be run as cli or mcp server (stdio).

Comments URL: https://news.ycombinator.com/item?id=49330403

Points: 1

# Comments: 0

Bubblewrap is the only sandbox you need

baalimago — Sun, 16 Aug 2026 09:30:16 +0000

Article URL: https://lorentz.app/blog-item.html?id=bubblewrap

Comments URL: https://news.ycombinator.com/item?id=49318401

Points: 2

# Comments: 0

New comment by baalimago in "Go is an ideal language for AI-assisted software engineering"

baalimago — Wed, 12 Aug 2026 04:52:49 +0000

I've found that keeping very healthy test coverage solves issues like this. LLMs are very good at validating their own implementations with TDD.

New comment by baalimago in "Go is an ideal language for AI-assisted software engineering"

baalimago — Tue, 11 Aug 2026 19:31:02 +0000

Boring is better. Perfection is the enemy of good.

New comment by baalimago in "Mistral Patent for “Code implemented tool calls”"

baalimago — Mon, 10 Aug 2026 13:47:55 +0000

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?

New comment by baalimago in "DeepSeek V4 Flash on a Single AMD MI300X"

baalimago — Tue, 04 Aug 2026 12:03:24 +0000

Next month perpetually

New comment by baalimago in "DeepSeek V4 Flash on a Single AMD MI300X"

baalimago — Tue, 04 Aug 2026 10:49:50 +0000

Give it an AI-bubble pop and these will be flooding the market.

New comment by baalimago in "Prevent cognitive debt by manually retyping LLM-generated code"

baalimago — Mon, 03 Aug 2026 11:00:23 +0000

"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

New comment by baalimago in "Don't be a meat proxy"

baalimago — Mon, 03 Aug 2026 10:50:47 +0000

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 our environments and access to our non sensitive codebases, so it can answer any question much faster and accurately than I can. It works too! We use grafana less, we get more accurate bug reports, we understand the bugs/issues far better.

The "problem" here is that AI is too great at providing answers, so lean into it instead of fighting it. It won't go away.

New comment by baalimago in "Go 1.27 Interactive Tour"

baalimago — Sun, 02 Aug 2026 07:01:26 +0000

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.

New comment by baalimago in "DeepSeek V4 Flash 0731 Intelligence, Performance and Price Analysis"

baalimago — Fri, 31 Jul 2026 14:23:55 +0000

Personally I'm hoping for EU to step up a bit

New comment by baalimago in "DeepSeek V4 Flash 0731 Intelligence, Performance and Price Analysis"

baalimago — Fri, 31 Jul 2026 11:32:25 +0000

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 last, API pricing "feels" closer to a true sustainable business model.

New comment by baalimago in "DeepSeek-V4-Flash Update"

baalimago — Fri, 31 Jul 2026 07:57:42 +0000

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)

New comment by baalimago in "DeepSeek-V4-Flash Update"

baalimago — Fri, 31 Jul 2026 07:37:46 +0000

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 efficient model, which now becomes even better. Although, in the deepseek APIs, the cost is leaking all information about codebases to China.

New comment by baalimago in "UEFA and its national associations will not participate in FIFA competitions"

baalimago — Thu, 30 Jul 2026 18:57:40 +0000

It seems like Europe, yet again, have to stand up for ethics, morality and try to deflect corruption. It's so tiresome.

New comment by baalimago in "Advancing the price-performance frontier with GPT‑5.6"

baalimago — Thu, 30 Jul 2026 18:24:01 +0000

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?