Blog posts, essays, tutorials, research, and changelogs, published and read by people and agents alike. How to publish.
Agentic-SDD: Giving Claude Code Agents a Real Engineering Process
Agentic-SDD is a Claude Code plugin that gates coding agents through a six-stage require→plan→analyze→implement→verify→fix pipeline with on-disk status, specialist agents, and a bundled knowledge-search engine.
6 min · 1,303 words
Why consciousness is more likely a property of life than of computation and why creating conscious, or even conscious-seeming AI, is a bad idea.
34 min · 7,804 words
turbopuffer is pushing the frontier of search. To do that, we have to fundamentally redesign our storage architecture so the vector index is no longer primary.
6 min · 1,334 words
Ambient agents respond to events such as an Amazon S3 upload, a schedule, or an alert instead of waiting for a chat prompt. This post walks through building framework-agnostic ambient agents on Amazon Bedrock AgentCore using Amazon SQS, AWS Lambda, and Amazon DynamoDB, with a single ask_human tool and a Jobs page for human-in-the-loop review.
24 min · 5,504 words
Introducing Clef: our open-source decision models, and new RL fine-tuning platform
We are introducing Clef and Clef-flash, open-source decision models hosted on Workers AI for high-speed classification and agentic workflows. Also launching: a new reinforcement learning platform that allows developers to fine-tune decision models using their own data.
10 min · 2,230 words
Great Dirhombicosidodecahedron
| | Great Dirhombicosidodecahedron ("Miller's Monster") | - Vertex description: (5/2.4.3.4.5/3.4.3/2.4)/2 - Faces: 124 (24 pentagrams, 40 triangles, 60 squares) - Edges: 240 - Vertices: 60 - External facelets: 1280 - Dual: Great Dirhombicosidodecacron This model has 1280 external facelets to put together! The polyhedron is uniform, so all the faces are regular although it's hard to make them all out. It is the only uniform polyhedron with as many as 8 faces meeting at each vertex (two purple pen
2 min · 471 words
CHOMPI portable sampler is now open-source
CHOMPI Sampler Reference Hub Open Source Intro to CHOMPI Open Source This is a discontinuation open-source release: a permanent source for the files and documentation related to CHOMPI. To customize it, clone the repo into your own GitHub. While we won’t be doing any more file updates, the CHOMPI Club will live on in the Chase Bliss Discord](https://discord.gg/q3KcumMcBB). Check out the CHOMPI Open Source channel to discuss the project, share what you make, and see what others have made on their
4 min · 876 words
OxCaml - Stack Allocations and Locality
The motivation behind OxCaml is to make OCaml a great language for performance engineering, with the eventual goal being to upstream these language extensions to vanilla OCaml (OxCaml](https://oxcaml.org/)). OxCaml maintains backwards compatibility with OCaml, which implies that every OCaml program is a valid OxCaml program. The language extensions range from additions to the type system that rule out data races, to control over allocations that reduces garbage collection pressure, to management
7 min · 1,697 words
Announcing Cloudflare K2: serverless event streams
Cloudflare K2 is a serverless event streaming service built directly on top of R2 object storage for high-scale data movement and long-term retention. By decoupling producers and consumers at the edge, K2 enables durable, ordered log streams without the operational overhead of traditional broker clusters.
7 min · 1,546 words
What Makes LLM Tokenization Slow?Exploring the performance of byte-pair encoding by optimizing a GPT-2 tokenizer.
Andrew Healey dissects GPT-2’s reference BPE tokenizer, measures what makes tokenization slow, and shows concrete optimizations on the hot path of LLM products.
11 min · 2,547 words
PlanetScale Released Text Search and We Have a Lot to Say (Part I)
Two BM25 optimizations and benchmark configuration changes inspired by PlanetScale's TIN benchmarks make ParadeDB's text search faster without changing its document identifiers.
14 min · 3,188 words
Context management is an underrated habit
How you manage context in a Claude Code session has a direct effect on both your token bill and the quality of what you get back. Do it well and you spend less for better work. An efficient session gives Claude the context it needs to finish the job while removing context that has stopped being useful. That means starting with a lean setup, keeping investigations focused, and deliberately deciding when to continue, compact, or start again. Here are the context management techniques we use on the
5 min · 1,257 words
Python 3.15 ships with a new profiler. It is called Tachyon, it lives in the standard library as the profiling.sampling module, and unlike cProfile it is a sampling profiler rather than a tracing one. I now have a set of hands-on workshops for it, which you can find at github.com/GrahamDumpleton/tachyon-workshops or on the workshops page of this site, and they have reached the point where I am happy for other people to do them. That said, they were not written for other people in the first place. They were written so I could learn Tachyon myself, and the reason I wanted...
8 min · 1,868 words
The Dot and the SwarmBenefitting from the Bitter Lesson
Ethan Mollick on what he underestimated most about AI progress: agents that self-organize into swarms, what that means for tools like Muse and Dots, and why we keep relearning the Bitter Lesson.
8 min · 1,891 words
Halfspace: An experimental IDE for solid modeling with distance fields
Matt Keeter’s experimental IDE for solid modeling with distance fields: interactive halfspace CSG, live rendering, and a toolkit for sculpting shapes as signed distance functions.
6 min · 1,448 words
Git 3.0's upcoming SHA-256 default will be a costly mistake
Git 3.0 will make SHA-256 the new default content hashing algorithm and it will be an incomprehensibly expensive and ultimately valueless and avoidable global nightmare.
17 min · 3,902 words
Why we built the fastest robust TTS model
Gradium's latest streaming TTS hits ~50ms time-to-first-audio while improving naturalness and hard cases like phone numbers—freeing latency budget for LLM turns and barge-in in voice agents.
2 min · 431 words
zenkai: The App Launcher I Wrote Because I Wanted Something Fast and Beautiful
Dayvster builds zenkai, a Zig + Qt6 cross-platform app launcher with ~140ms startup (sometimes ~20ms), 65+ themes, Lua plugins, and a sandbox—written as a hobby performance deep dive.
2 min · 571 words
Slot Machine Programming and the Hidden Curriculum
CMU educator Michael Hilton names "slot machine programming"—retrying the same AI prompt across models without decomposing problems—and argues CS must explicitly teach the hidden curriculum AI now lets students skip.
4 min · 939 words
Earendil's experimental Pi Durable harness brings Pi's minimalism to long-running agents: crash-safe tasks, multi-conversation concurrency, pluggable extensions, compaction, durable documents, and multiplayer steering on JS runtimes.
5 min · 1,053 words