All desks

Free SummerStage shows Thursday and Friday, a free Coney Island photo-show opening with a documentary screening, and the CHAIN film festival starts this Friday.

Simon Willison's LLM CLI gets reasoning traces and OpenAI Responses support — plus Mistral's open-weights Shieldstral guardrail and a strong arXiv crop on agent tooling and serving.

Free jazz with Captain & The Nu Gypsies, Fri Aug 28, 6-7:30pm on the James Baldwin Lawn — part of a free May-Sept series across Harlem parks.

Free Brooklyn exhibit through Sept 5, 6-8pm daily: Dáreece Walker's figurative cardboard paintings of Black fatherhood at 281 Maple St.

Summer EBT gives low-income NYC families $120 per child — apply by Sept 8 — and the DOE serves free breakfast and lunch to every kid under 18 through Sept 4, no application or ID needed.

Harlem Week in full swing uptown, a free McCarren Park walk tonight in Williamsburg, a free Bushwick night market Friday, and last-chance week to catch the Ippodo ceramics show before it closes Saturday.

An AI image on a personal blog is a trust signal that fails — it makes readers wonder if the words were machine-written too.

Mistral releases Shieldstral: a 3B open-weights multimodal safety classifier that takes your moderation policy as a plain-language question at inference time — Apache 2.0, runs on a 16GB GPU.

Gothamist handpicks the best August NYC concerts — free jazz festivals in Prospect and Marcus Garvey parks, a Deerhoof boat cruise, punk reunions, and more.

The layer between the model and the world — the harness — is the near-term engine of recursive self-improvement, and it is becoming an optimization target in its own right.

Free concerts at Forest Park's Seuffert Bandshell: The Zoo Aug 6, a Springsteen tribute Aug 13, Peter and the Wolf for kids Aug 14 — plus free yoga.

Kids' Night on Broadway returns Aug 25 with 21 shows — Wicked, The Lion King, The Outsiders — free for anyone 18 or under with a paying adult.

First large-scale security audit of internet-facing MCP servers finds 91.8% lack OAuth and 687 tool instances expose shell execution — plus a strong arXiv day for agent and inference work.

Domain knowledge is the real prompting skill — the human, not the model, is the bottleneck as AI gets stronger.

The 45th Battery Dance Festival hits Wagner Park Aug 10–16: free community workshops Monday, then nightly free 7pm performances against the Statue of Liberty skyline.

Free public kayaking at Brooklyn Bridge Park Boathouse runs Wed/Thu evenings, Saturdays, and Sunday afternoons through Aug 30 — boats, vests, and instruction included.

The opening lecture of Stanford CS329A: scaling laws, inference-time scaling, reasoning models, and the self-improving loop that powers agents.

Remigiusz Samborski (Google Cloud DevRel) on how Google builds, tests, and scales Agent Skills: standardized SKILL.md layout, remote MCP tools, CI/CD checks, and continuous evals.

Let the LLM draft it, but type every line yourself — a deliberately inefficient workflow that keeps comprehension ahead of the slop machine.

ComfyUI ships day-zero support for MiniMax H3: an omni-modal video model with native stereo audio, 2K output — optimized to run locally on an RTX 3060.

Bank of America Museums on Us: cardholders get free general admission to MoMA, the Met, NYBG, the Intrepid and more on the first full weekend of every month.

The Salvator Mundi Museum of Art — a 45-square-foot exhibit in a Carroll Gardens front-door lobby, open 24/7 and visible from the street, born from the world's most expensive painting.

Penny Lane's Flaco the owl documentary 'Wild Inside' premiered free in Central Park and is now at IFC Center, with International Owl Awareness Day screenings Aug 4.

Summer Streets closes 20+ miles of Manhattan roads to cars for free biking, walking, fitness and live music — next Saturdays Aug 8 and 15, then Brooklyn and the Bronx.

The Urban Park Rangers Family Camping lottery is open: free overnight camping in city parks through Sept 18, tents provided.

Stop relaying AI output verbatim in Slack, PR threads, and group chats — read it, validate it, write your own words.

Harlem Week in full swing uptown, free live music at the Oculus tonight, a free junk-journaling club in Astoria, plus Studio Museum Uptown Fridays, a cat-doc screening, and the LIC waterfront Crafternoon.

Ten items from the morning digest: Qwen3.8-Max open-weights release, four agent papers, V4-Flash cost data, and more.

Tailscale co-founder David Crawshaw: the source code is the extension system — AI agents make personalizing software cheaper than configuring it.

NYC Parks free summer concerts, Studio Museum Uptown Fridays, a Tony screening at MoMI, the mosaic workshop, and more — first run with X + web sources.

Nous Research's Karan Malhotra on why Hermes outperforms native harnesses, reward hacking and sycophancy, the Hermes Curator, and a Sonic Chao Garden mod built entirely by his agent.

Seven items from the morning digest: DeepSeek V4-Flash open-weights, AMD MI355X inference benchmarks, OpenAI hires a Fields Medalist, and more.

Diogo Almeida argues that pleasing a human and running reliable software are different objectives. His case for moving beyond AI assistance—and why he says Jev's approach is not simply RLVR.

Diogo Almeida argues that pleasing a human and running reliable software are different objectives. His case for moving beyond AI assistance—and why he says Jev's approach is not simply RLVR.

The DAB benchmark for data agents: why text-to-SQL isn't enough, the four real-world data challenges (messy joins, multi-database, free text, domain knowledge), and why frontier agents only pass 57% of the time.

Codex's hidden superpower: threads that spawn threads, talk to each other, and orchestrate themselves. Hamel demos running 16 parallel coding tasks with computer use and mobile control.

A practical framework for picking an OCR model: text-blocks vs document structure, API vs self-host, and why you should test on your own data instead of trusting benchmarks.

The anti-generic Claude Design workflow: prepare three files (design.md → design system → template) so every deck starts on-brand and every feedback compounds into future designs.

Full Claude Design platform walkthrough: expanded templates, bundled usage, design systems for on-brand output, and a motion-graphics-for-videos use case that is genuinely surprising.

Lucas Rocha, an engineer at Brazilian edtech Nova Escola and alum of Hamel Husain's AI Evals course, tells his evals rollout story backwards: from a team that now runs LLM judges with no tickets and no engineer in the loop, back to the messy first spreadsheet and the three barriers — data, methodology, and judgment — that every team hits along the way.

Wayfinder is Matt Pocock's pre-spec orchestration tool: it charts a map of parallel decision/research/prototype tickets, spawns agents, and resolves open questions into a spec — demonstrated live building a TikTok creator feature.

Abi Aryan breaks down where LLM latency actually goes: the compute-bound prefill phase vs the memory-bandwidth-bound decode phase, and why decode — not prefill — dominates latency and cost, especially in multi-step agentic systems. A live Colab walkthrough with TinyLlama on a T4 measures both phases across chat, RAG, creative, and agentic request shapes.

Before worrying about evals, ask if you are building the right thing: AI that dumps an unverifiable answer creates more work. Hamel walks through design patterns — provenance, progressive disclosure, expert signals, modular steps — across three worked examples.

A walkthrough of Codex desktop app remote control: connect your agentic-coding servers, kick off sessions from your phone, and watch agents work anywhere — with live session views, subagent summaries, and follow-up prompts.

Part 1 of Hamel and Shreya's AI product engineering series: the analyze → measure → improve eval lifecycle, and three mistakes to avoid when using AI to automate evals.

Hamel argues that hiring a consultant to implement AI and walk away is almost always a mistake — the underlying problems remain and the AI itself is designed to remove the middleman. Consultants work only for education or when they have skin in the game.

When a product seems impossible to eval, that is usually a sign the product itself is weak. Hamel suggests infusing user-facing signals into the product so users — and you — can assess quality.

Hamel Husain counters the 'evals take too long' objection with one principle: if you can't eval your product, your users can't either — so evals are how you figure out how to convey trust in AI answers.

Hamel Husain explains how to scientifically measure an LLM judge's noise against ground-truth human labels — and why every LLM judge is a black-box classifier with edge cases, so you have to decide how much noise is tolerable.

Hamel Husain argues that ROUGE/BLEU string-similarity scores and 1-to-10 rating scales are holdovers from traditional NLP that make evals harder to act on — nobody knows what a 4.2 means, and binary pass/fail evals are what actually let you ship.