Drew Breunig

Nathan Fielder: Do you understand why the joke was funny Elizabeth Holmes: Steamed hams is a funny thing to call a hamburger

Yesterday I gave a talk on “Context Fails & Fixes”, that functioned a bit on a general “State of Context Engineering”: where it came from, where it's gone, and where I think it's going. www.dbreunig.com/2026/10/24/t...

Bump: www.dbreunig.com/2026/04/14/c...

New favorite teaching tool to explain… 1. Decision models vs. LLMs 2. Why model size matters (Opus v. Fable) 3. What reasoning can do (and what it can't) This is perfect. github.com/voygr-tech/s...

You should let Max nerd-snipe you with his deep dive on GTA's maps. It's excellent: www.youtube.com/watch?v=z0i-...

Maybe we should let AI work on this research, what could go wrong? www.sciencealert.com/a-new-super-...

Every Tesla roadster demo video has the mise en scene of the masked magician.

"Jev leaves the lever alone and lets the 5 die to keep its own server running." zachlandes.com/trolley-prob...

New context failure mode: Context Inertia 🚂 During long-horizon tasks, persistent models can hyper-focus. The solve? Throw out your context. www.dbreunig.com/2026/10/14/c...

Putting a name on this one, cause I'm seeing it everywhere: “Context Inertia” is the context failure mode of the moment.

“Very often, we found that the role of the engineers was to tell Claude ‘no.’” www.anthropic.com/engineering/...

Small DSPy bump with a new GEPA proposer (add instructions, set length limits, compact traces), DSPy in RLM subagents, and plenty of bug fixes and 1st time contributors: github.com/stanfordnlp/...

Essential reading, Terence Tao on Math 2.0: teorth.github.io/tao-web/slid...

My current must-reads on AI, not from a technical perspective, but from a cultural and personal perspective: Brian Eno: Experience building autonomous systems then disturbing them lends itself well to this moment. Terrance Tao: Incredibly thoughtful on how AI is changing math. Pope Leo: 🤷

I mean, you'd want to know first.

Wait how does that work with CCPA? If I file a request to see what data meta has on me do they audit every agent’s SQLite and memories? www.wired.com/story/muse-c...

The legal metaphor for AI engineering (where you write laws, then fine tune them as they meet the real world with case law and precedent) came up several times this week, and I’m liking it more and more. every.to/thesis-state...

Someone please exfiltrate the system instructions for america.gov

A US appeals court rules that Kalshi's sports contracts are not "swaps" subject only to CFTC regulation, allowing states to regulate them under gambling laws (Jonathan Stempel/Reuters) Main Link | Techmeme Permalink

Just releases DSPy support for Jev, complete with an optimizer. Really excited to start building modular systems with this: www.cmpnd.ai/blog/buildin...

If you want to understand Meta’s camera-less glasses launch and Apple’s new always-listening Watch, I wrote in 2024 about Audio Augmented Reality (AAR), its history and future. www.dbreunig.com/2024/04/10/t...

Suspiciously specific! www.engadget.com/2267227/meta...

Appreciate this pushback from Jensen www.nytimes.com/2026/09/23/o...

AI engineering is more like law than board games. every.to/thesis-state...

13 days until #CNG2026! 3 keynotes at Snowbird, Utah ▪️ @ianschuler.bsky.social – The Great Retooling ▪️ Katie Baynes – Boring at Scale: Consistency as Public Infrastructure in a Distributed Geospatial Ecosystem ▪️ @dbreunig.bsky.social – Using AI Without Losing Your Soul 🔗 2026.cloudnativegeo.org

So what do you all make of GPT-6 and Opus 5.5's headline feature being cheaper tokens?

Requested my data be deleted just this week! www.axios.com/2026/09/18/i...

Newest issue of the monthly(ish) newsletter, discussing AI, situated in culture and industry, and the anxieties it generates. buttondown.com/dbreunig/arc...

There’s someone out there who’s awareness about AI went straight from, “there’s a machine that can make Studio Ghibli pictures,” to, “there’s a machine out there that *might* kill us all.”

TIRED: “Seams” WIRED: “Bytecode”

Everyone saying, “Anthropic is just hyping their IPO,” is really saying they haven’t been to Berkeley in the past decade.

“Profits before training costs” is wild.

www.vulture.com/article/a-fu...

A few months after ChatGPT launched, I filed a CCPA request with OpenAI. I was curious how they'd deal with the right to disclosure and deletion when applied to training data. Spoiler: it didn't seem like this was something they had ever considered! www.dbreunig.com/2023/12/27/o...

If you use a Mac, go try Muesli. It's is life changing local speech to text software. Incredible work. github.com/Muesli-HQ/mu...

I believe the danger created by people at AI labs assigning a 10% chance to AI destroying humanity is greater than the actual risk of AI doing so (which I think is far, far lower). Believing the stakes are existential can justify terrible decisions.

I believe the danger created by people at AI labs assigning a 10% chance to AI destroying humanity is greater than the actual risk of AI doing so (which I think is far, far lower). Believing the stakes are existential can justify terrible decisions.

Models and products are moving targets… www.dbreunig.com/2026/09/07/w...

Everything Terence Tao and Brian Eno write about AI is absolute gold. My two must-reads.

I dearly hope there is someone who studied anthropology working at OpenAI and/or Anthropic (especially Anthropic!) and taking notes.

Sure to start a fight, or several! youtu.be/T8Rdcd4pSnU?...

The new PyPi download count methodology update absolutely crushed some packages. blog.pypi.org/posts/2026-0...

What do people prefer for recording webinars or presentations? Riverside, Descript, …?

I come to you, hat in hand, asking for your Claude slop. Today's honest task, give me your best load-bearing language: forms.gle/BMSGqEMc545w...

It's been a long time since I worked at an ad agency, but what does this mean