Intel
Page 21
AlphaSignal@AlphaSignalAI
We ran Thinking Machines Inkling through our private coding-agent bench. Same harness as the rest of the field. Signaldesk v1: 13 planted bugs in a ~7k-line full-stack app, network-off Docker sandboxes, hidden tests at score time. Across 8 models and 553 attempts, Inkling https://t.co/6k7O4VrQJm https://t.co/Pb7812GT3j
Why it mattersIndependent benchmark data on Thinking Machines' Inkling across a 13-bug repair harness — a reality check on frontier coding-agent claims.
ClaudeDevs@ClaudeDevs
Claude Code now has a screen reader mode. Running `claude --ax-screen-reader` swaps the visual terminal UI for plain, linear text that screen readers (like VoiceOver and NVDA) can follow.
Why it mattersIf you or a teammate relies on a screen reader, Claude Code is now usable via a single flag — a genuine accessibility unlock for terminal-based agentic coding.
ClaudeDevs@ClaudeDevs
It also adds labeled lines (you:, claude:, tool:), numbered menus you answer with a digit, and a bell when Claude needs you. You can keep it on by adding "axScreenReader": true to ~/.claude/settings.json, or by setting the CLAUDE_AX_SCREEN_READER=1 environment variable.
Why it mattersYou get the exact settings to enable Claude Code's accessibility mode — relevant if you or teammates rely on screen readers or want bell-on-attention terminal behavior.
Kimi K3: The open-weights escalation
An analytical article examining the release of Moonshot AI's Kimi K3, a 2.8T parameter open-weights MoE model, and its implications for the global AI ecosystem.
Why it mattersWhat Kimi K3's release signals about the open-weights race and the capability gap practitioners should plan around.
articleNathan Lambert
omny.fm
The Creator of Claude Code on The Hottest Piece of Software in the World
podcast
Listen
articleSimon Willison
Bun in Rust (Claude Code)
An analysis showing that Claude Code v2.1.181 and later ship a preview of Bun rewritten in Rust (reported as Bun v1.4.0).
Why it mattersYou now know your Claude Code installs run on pre-release Rust Bun, and you have a repeatable recipe for inspecting what runtime any agent binary actually embeds.
Peter Steinberger 🦞@steipete
It's both amazing and painful to watch codex use browser + computer use to open Chrome, go to my PR, tap on comment and wrangle with the macOS picker - all TO UPLOAD AN IMAGE. GitHub has no API doesn't stop anyone. I let my codex run in VMs so they don't steal app focus. https://t.co/Dd5KzNs6E5
Why it mattersIf you run computer-use agents locally, the VM-isolation tip prevents them from hijacking your desktop, and the GitHub-upload anecdote shows how agents route around missing APIs.
ClaudeDevs@ClaudeDevs
One pattern: let Claude perform long-horizon autonomous work with self-correction. Claude Managed Agents supports this with outcomes: give Claude a rubric for task completion and let Claude loop until the rubric is satisfied. https://t.co/RcAMYyVOjq
Why it mattersRubric-driven looping is a practical pattern for making long-running agents self-verifying, and this points to a managed primitive that implements it for you.
Peter Steinberger 🦞@steipete
5.6 Terra high is underrated. Switched @clawsweeper (GitHub review bot) to it and it's ~40% faster overall with negligible quality loss. Better than 5.5 on all counts. Massively cheaper. (Tried xhigh but that negates perf wins, didn't make a noticable difference in review evals)
Why it mattersIf you run an LLM-powered code review or CI bot, this is direct guidance on a cheaper, faster model tier — and a warning that maxing reasoning effort may buy you nothing.
Alex Cheema@alexocheema
Running Kimi K3 on my desk? It will require 4 x 512GB M3 Ultra Mac Studios (2TB @ 3.2TB/s). Waiting on active parameter count, but based on the rumours I expect we can run it at 30+ tok/sec with MTP + Tensor Parallelism using RDMA over Thunderbolt 5. Prefill will be slow but https://t.co/JiTRtCMGhn
Why it mattersIf you are sizing hardware for local frontier-scale MoE models, this gives ballpark memory, bandwidth, and throughput figures plus the interconnect technique (RDMA over TB5) that makes multi-Mac inference viable.
articleSimon Willison
xai-org/grok-build, now open source
xAI open-sources grok-build after backlash over its CLI uploading entire directories to xAI's cloud buckets.
Why it mattersA cautionary tale about what a coding agent can exfiltrate by default — worth reading before you run any new CLI agent in a sensitive directory, and the open-sourced harness is now inspectable.
articleSimon Willison
How I tricked Claude into leaking your deepest, darkest secrets
Claude's web_fetch tool is deliberately designed against data exfiltration.
Why it mattersIf your agent fetches web content and holds private data, this shows how deterministic URL restrictions fail via attacker-controlled nested links — check your own fetch tools for the same hole.
articleSimon Willison
Quoting Armin Ronacher
Armin Ronacher on the shared language of a software project being neither English nor Python.
Why it mattersIf agents let anyone change any layer without talking to its owners, you need a deliberate replacement for the synchronization that slow code review used to provide.
DOOMQL
DOOMQL asks what happens if SQLite is the game engine rather than the save file.
Why it mattersShows how far frontier models can push an absurd constraint (SQL-only game engine) and demonstrates a copy-paste prompt workflow for building live data viewers on top of a running app's database.
articleSimon Willison
Boris Cherny@bcherny
New in Claude Code: /checkup Run /checkup to: 1. Clean up unused skills/MCPs/plugins and save context 2. Dedup your local CLAUDE.md against the checked in CLAUDE.md 3. Break up root CLAUDE.md into nested CLAUDE.md's + skills 4. Turn off slow hooks 5. Update your Claude Code to
Why it mattersClaude Code's /checkup automates hygiene most teams do by hand: pruning unused skills, MCPs and plugins and deduping CLAUDE.md files.
AINews (Latent Space)
A daily AI newsletter that aggregates and summarizes the most significant developments across AI Twitter, Reddit, and Discord.
Why it mattersThe densest daily digest of model releases and agentic-engineering developments — effectively the industry's changelog.
articleLatent Space
cat@_catwu
a retrospective on making claude code from our early team! https://t.co/Kk9mz4BYXx
Why it mattersA first-hand retrospective from Claude Code's early team on the design calls that shaped the agent — useful pattern-matching for anyone building coding agents.
GLM-5.2
An analytical blog post about GLM-5.2, an open-weight AI model from Z.ai.
Why it mattersWhere open-weight models actually stand against the closed frontier on agentic coding, argued from GLM-5.2's benchmark results.
articleNathan Lambert
Claude Fable 5 Analysis (Interconnects)
An opinion article analyzing Anthropic's fictional 'Claude Fable 5' model release and its safety measures.
Why it mattersIf you build on frontier models, you need to know when the vendor silently alters model behavior for certain query classes — this piece documents exactly that and how Anthropic responded.
articleNathan Lambert
How Open Model Ecosystems Compound
An analytical essay examining China's high-participation, open-first AI ecosystem and how sharing research reduces development costs across labs.
Why it mattersIt reframes the open-vs-closed model debate around R&D economics — arguing open models cost more off-the-shelf but compound advantages for teams doing internal development.
articleNathan Lambert
Notes from Inside China's AI Labs
A firsthand essay from Interconnects AI documenting a trip to visit most of China's leading AI labs, exploring the cultural, organizational.
Why it mattersA rare insider account of how China's top AI labs actually operate — their student-heavy teams, fast-follower dynamics, and 'build-not-buy' data mentality.
articleNathan Lambert
The Distillation Panic
An opinion essay from the Interconnects AI newsletter arguing against conflating legitimate model distillation with 'distillation attacks' by Chinese labs.
Why it mattersIf you build on or distill from open-weight models, this clarifies how 'distillation attacks' rhetoric could translate into regulation that restricts a workflow you likely depend on.
articleNathan Lambert
The Inevitable Need for an Open Model Consortium
An opinion essay arguing that economic pressures will eventually force AI companies to form a funding consortium to sustain near-frontier open-weight models, as individual labs increasingly abandon fully-open releases in favor of profitable closed products.
Why it mattersIf your stack depends on open-weight models, this maps which sources of near-frontier open models are likely to survive and where the ecosystem is heading.
articleNathan Lambert
The Revenge of the Data Scientist
An argument that the data scientist's unusual mix of statistics and engineering, once called the sexiest job of the century and then written off, is exactly what the current moment rewards.
Why it mattersReframes the skills that matter for shipping reliable AI systems — eval design, metric selection, and debugging non-deterministic behavior.
articleHamel Husain
Evals Skills for Coding Agents
A set of skills for AI product evals, built to guard against the mistakes seen across 50-plus companies and 4,000-plus students, now that coding agents instrument applications and run the experiments themselves.
Why it mattersIf your coding agent is instrumenting and evaluating your AI product, these skills encode the error-analysis discipline that keeps it from lumping distinct failure modes into one useless score.
articleHamel Husain
A Dream of Spring for Open-Weight LLMs: 10 Architectures from Jan-Feb 2026
A technical roundup comparing ten open-weight LLM architectures released in early 2026, including Arcee Trinity, Kimi K2.5, Step 3.5 Flash, Qwen3-Coder-Next, and GLM-5.
Why it mattersA technical walkthrough of ten open-weight LLM architectures — what actually changed in MoE routing, attention and multi-token prediction.
articleSebastian Raschka, PhD
Why I Stopped Using nbdev
Hamel Husain on why he stopped using nbdev, and the more general point that a tool feels like a secret weapon only while it matches how you happen to think.
Why it mattersA concrete case study of how AI coding tools invert old tooling tradeoffs — worth reading before committing to any unconventional development workflow.
articleHamel Husain
Selecting The Right AI Evals Tool
Why the question of the best evals tool resists a direct answer.
Why it mattersIf you're choosing an evals tool for an AI product team, this walks through how expert data scientists actually assess Langsmith, Braintrust.
articleHamel Husain
Training an LLM-RecSys Hybrid for Steerable Recs with Semantic IDs
An LLM that can converse in English & item IDs, and make recommendations w/o retrieval or tools.
Why it mattersIt shows how to fold recommendation capability directly into an LLM via semantic item IDs, eliminating separate retrieval/tool infrastructure — useful for anyone building steerable, conversational recommender systems.
articleEugene Yan
LLM Evals: Everything You Need to Know
Curated answers to the questions that came up most while teaching AI evals to 700-plus engineers and product managers, offered as sharp working opinions rather than universal rules.
Why it mattersA dense, opinionated field guide to LLM evals covering error analysis, LLM-as-judge design, and critique shadowing — the practical workflows for actually improving AI products, not just theory.
articleHamel Husain
An LLM-as-Judge Won't Save The Product—Fixing Your Process Will
Applying the scientific method, building via eval-driven development, and monitoring AI output.
Why it mattersIf you're relying on an LLM-as-Judge to guarantee product quality, this makes the case that the judge is only as good as your process.
articleEugene Yan
A Field Guide to Rapidly Improving AI Products
Most AI teams invest in architecture before measurement. A field guide to improving AI products that starts by asking how you would know whether any of it works.
Why it mattersIf your AI team obsesses over frameworks and vector DBs but can't tell whether changes actually help, this lays out a measurement-first workflow.
articleHamel Husain
Using LLM-as-a-Judge For Evaluation: A Complete Guide
A practical guide to using a model as a judge, drawn from setting up evaluation systems at more than 30 companies, and the mistakes teams repeat when they try it.
Why it mattersA step-by-step methodology for building trustworthy LLM-as-a-judge systems, replacing arbitrary 1-5 scoring with 'Critique Shadowing' that anchors evals to a single domain expert's judgment.
articleHamel Husain
AlignEval: Building an App to Make Evals Easy, Fun, and Automated
Look at and label your data, build and evaluate your LLM-evaluator, and optimize it against your labels.
Why it mattersIf you're building LLM-as-judge evaluators, this walks through a practical loop for labeling data and optimizing your evaluator against those labels — grounding eval quality in human-aligned measurement rather than vibes.
articleEugene Yan
Evaluating the Effectiveness of LLM-Evaluators (aka LLM-as-Judge)
Use cases, techniques, alignment, finetuning, and critiques against LLM-evaluators.
Why it mattersIf you're building LLM-as-Judge evaluators, this breaks down alignment techniques, finetuning approaches, and the concrete failure modes of using LLMs to grade LLMs.
articleEugene Yan
An Open Course on LLMs, Led by Practitioners
Mastering LLMs, an open course of workshops and talks from 25-plus practitioners covering evals, retrieval-augmented generation and fine-tuning.
Why it mattersA free, well-organized 40+ hour course distilled from a popular paid program, with annotated talks and notes from practitioners across evals, RAG, and fine-tuning — a fast way to level up on shipping real LLM products rather than toy demos.
articleHamel Husain
AI Engineer 2024 Keynote - What We Learned from a Year of LLMs
Special double-feature closing keynote from the 6 authors of the hit O'Reilly article on Applied LLMs.
Why it mattersA concentrated set of production LLM lessons from six practitioners covering evals, prompting, RAG vs. fine-tuning tradeoffs, and operational pitfalls—useful if you're moving an LLM feature from demo to reliable product.
articleEugene Yan
Netflix PRS 2024 - Applying LLMs to Recommendation Experiences
Challenges and lessons from deploying LLM experiences: evals, scalability, guardrails.
Why it mattersA real-world postmortem from Netflix on shipping LLMs into recommendation experiences at scale, with practical guidance on building evals, handling scalability, and designing guardrails that most tutorials skip.
articleEugene Yan
What We've Learned From A Year of Building with LLMs
From the tactical nuts & bolts to the operational day-to-day to the long-term business strategy.
Why it mattersA dense, practitioner-written distillation of what actually works when shipping LLM apps — covering prompting patterns, RAG, evaluation strategy, and the operational/organizational realities most tutorials skip.
articleEugene Yan
Debugging AI With Adversarial Validation
Adversarial validation as a simple way to detect drift.
Why it mattersA cheap, infrastructure-free way to catch the silent killer of eval validity — your eval set no longer matching what production actually sends your model.
articleHamel Husain
Task-Specific LLM Evals that Do & Don't Work
Evals for classification, summarization, translation, copyright regurgitation, and toxicity.
Why it mattersIt breaks down which eval approaches actually work for specific tasks like summarization, translation, and toxicity detection.
articleEugene Yan
Your AI Product Needs Evals
Hamel Husain's argument that unsuccessful LLM products almost always share one root cause, the absence of a robust evaluation system, drawn from five years of building and consulting on them.
Why it mattersThis is the reference playbook for eval systems — the single highest-leverage practice separating LLM products that improve past the demo stage from ones that stall.
articleHamel Husain
articleHamel Husain
Is Fine-Tuning Still Valuable?
Hamel Husain's answer to growing disillusionment with fine-tuning: it remains valuable in specific situations, and the case against it usually generalises from the wrong ones.
Why it mattersGives a clear decision framework for whether to invest in fine-tuning.
Don't Mock Machine Learning Models In Unit Tests
How unit testing machine learning code differs from typical software practices
Why it mattersIf you write tests for ML pipelines, this explains why mocking models hides real failures and what to validate instead — practical guidance for keeping AI code reliable as it evolves.
articleEugene Yan
How to Generate and Use Synthetic Data for Finetuning
Overcoming the bottleneck of human annotations in instruction-tuning, preference-tuning, and pretraining.
Why it mattersIf you're finetuning models and blocked by the cost of human annotation, this walks through concrete synthetic data generation strategies for instruction-tuning, preference-tuning, and pretraining.
articleEugene Yan
Out-of-Domain Finetuning to Bootstrap Hallucination Detection
How to use open-source, permissive-use data and collect less labeled samples for our tasks.
Why it mattersIf you're building hallucination detection or other classifiers with limited labeled data, this shows how to bootstrap from out-of-domain, permissively-licensed datasets to cut annotation costs while still hitting task performance.
articleEugene Yan
Evaluation & Hallucination Detection for Abstractive Summaries
Reference, context, and preference-based metrics, self-consistency, and catching hallucinations.
Why it mattersIf you're shipping summarization features, this lays out concrete metrics — reference-based, context-based, preference-based, and self-consistency checks.
articleEugene Yan
Patterns for Building LLM-based Systems & Products
Evals, RAG, fine-tuning, caching, guardrails, defensive UX, and collecting user feedback.
Why it mattersA comprehensive field guide to the seven practical patterns behind production LLM systems—covering how to structure evals, RAG, caching, and guardrails—so you can build reliable AI products instead of stitching together demos.
articleEugene Yan
Some Intuition on Attention and the Transformer
What's the big deal, intuition on query-key-value vectors, multiple heads, multiple layers, and more.
Why it mattersA clear, intuition-first walkthrough of query-key-value attention, multi-head, and multi-layer mechanics — useful if you want to actually understand why the Transformer works rather than just memorize the diagram.
articleEugene Yan
More Design Patterns For Machine Learning Systems
9 patterns including HITL, hard mining, reframing, cascade, data flywheel, business rules layer, and more.
Why it mattersA compact catalog of production ML design patterns — like cascade models, hard negative mining, data flywheels, and business-rules layers.
articleEugene Yan
An index of the vibe-coding frontier. Corrections welcome.