Hacker News: dbreunig

Claude Code builds a system prompt

dbreunig — Sun, 05 Apr 2026 14:47:21 +0000

Article URL: https://www.dbreunig.com/2026/04/04/how-claude-code-builds-a-system-prompt.html

Comments URL: https://news.ycombinator.com/item?id=47650000

Points: 2

# Comments: 0

The Cathedral, the Bazaar, and the Winchester Mystery House

dbreunig — Fri, 03 Apr 2026 17:01:52 +0000

Article URL: https://www.oreilly.com/radar/the-cathedral-the-bazaar-and-the-winchester-mystery-house/

Comments URL: https://news.ycombinator.com/item?id=47629178

Points: 2

# Comments: 0

The 2nd phase of OSS in the agentic era: From clones to reimaginings

dbreunig — Thu, 02 Apr 2026 16:17:48 +0000

Article URL: https://www.dbreunig.com/2026/04/01/the-2nd-phase-of-agentic-development.html

Comments URL: https://news.ycombinator.com/item?id=47616459

Points: 2

# Comments: 0

The Cathedral, the Bazaar, and the Winchester Mystery House

dbreunig — Wed, 01 Apr 2026 14:13:34 +0000

Article URL: https://www.dbreunig.com/2026/03/26/winchester-mystery-house.html

Comments URL: https://news.ycombinator.com/item?id=47601194

Points: 183

# Comments: 65

New comment by dbreunig in "If DSPy is so great, why isn't anyone using it?"

dbreunig — Mon, 23 Mar 2026 16:41:07 +0000

Model testing and swapping is one of the surprises people really appreciate DSPy for.

You're right: prompts are overfit to models. You can't just change the provider or target and know that you're giving it a fair shake. But if you have eval data and have been using a prompt optimizer with DSPy, you can try models with the one-line change followed by rerunning the prompt optimizer.

Dropbox just published a case study where they talk about this:

> At the same time, this experiment reinforced another benefit of the approach: iteration speed. Although gemma-3-12b was ultimately too weak for our highest-quality production judge paths, DSPy allowed us to reach that conclusion quickly and with measurable evidence. Instead of prolonged debate or manual trial and error, we could test the model directly against our evaluation framework and make a confident decision.

https://dropbox.tech/machine-learning/optimizing-dropbox-das...

Build a deep researcher and learn DSPy Signatures and Modules

dbreunig — Tue, 10 Mar 2026 23:25:11 +0000

Article URL: https://www.cmpnd.ai/blog/learn-dspy-deep-research.html

Comments URL: https://news.ycombinator.com/item?id=47330068

Points: 2

# Comments: 0

Can chat bots accommodate advertising?

dbreunig — Fri, 06 Mar 2026 18:29:17 +0000

Article URL: https://www.dbreunig.com/2025/09/02/considering-ad-models-for-ai-products.html

Comments URL: https://news.ycombinator.com/item?id=47279013

Points: 2

# Comments: 0

New comment by dbreunig in "Learnings from a No-Code Lib: Keep the Spec Driven Development Triangle in Sync"

dbreunig — Wed, 04 Mar 2026 21:23:13 +0000

No reason it can't. I know people currently generating specs from existing code; just gotta write the pipeline.

Learnings from a No-Code Lib: Keep the Spec Driven Development Triangle in Sync

dbreunig — Wed, 04 Mar 2026 18:39:08 +0000

Article URL: https://www.dbreunig.com/2026/03/04/the-spec-driven-development-triangle.html

Comments URL: https://news.ycombinator.com/item?id=47251886

Points: 4

# Comments: 3

New comment by dbreunig in "We gave terabytes of CI logs to an LLM"

dbreunig — Tue, 03 Mar 2026 16:13:31 +0000

"Think step by step," was just a sentence you appended to your prompt.

It ended up kicking off reasoning training which enabled the massive gains in coding, tool use, and more over the last 18 months.

So yeah, it's "just using LLMs in a specific way."

New comment by dbreunig in "Meta’s AI smart glasses and data privacy concerns"

dbreunig — Tue, 03 Mar 2026 04:36:35 +0000

Last year they pushed out an update stating if any “Meta AI” is left on, they can access image data for training,

I turned the AI off and used them as headphones and taking videos while biking. After a couple rides, I couldn’t bring myself to put them on because people started to recognize them and I realized I didn’t want to be associated with them (people are right to assume Meta has access to what they see).

Meta Ray Bans, if kept simple, could have been a great product. They ruined them.

Claude and the Dow: AI is unlike other tech because AI has embedded judgment

dbreunig — Mon, 02 Mar 2026 00:53:15 +0000

Article URL: https://www.dbreunig.com/2026/03/01/the-issue-is-embedded-judgment.html

Comments URL: https://news.ycombinator.com/item?id=47212545

Points: 1

# Comments: 1

New comment by dbreunig in "We gave terabytes of CI logs to an LLM"

dbreunig — Fri, 27 Feb 2026 16:10:59 +0000

Check out “Recursive Language Models”, or RLMs.

I believe this method works well because it turns a long context problem (hard for LLMs) into a coding and reasoning problem (much better!). You’re leveraging the last 18 months of coding RL by changing you scaffold.

Two Beliefs About Coding Agents: Devs Don't Realize What They Bring

dbreunig — Thu, 26 Feb 2026 01:19:45 +0000

Article URL: https://www.dbreunig.com/2026/02/25/two-things-i-believe-about-coding-agents.html

Comments URL: https://news.ycombinator.com/item?id=47160571

Points: 2

# Comments: 0

New comment by dbreunig in "Why is Claude an Electron app?"

dbreunig — Sun, 22 Feb 2026 01:12:38 +0000

Author of the post here.

I didn’t say AI was bad and I acknowledged the benefits of Electron and why it makes sense to choose it.

With 64gb of RAM on my Mac Studio, Claude desktop is still slow! Good Electron apps exist, it’s just an interesting note give recent spec driven development discussion.