Link radar, September 15, 2026
Five reads: Anthropic's CI scaling lessons, Copilot auto-model tiers, Cloudflare per-Worker access, Cloudflare search-without-training controls, and ShadowPEFT in PEFT.
2 min read↗
Ideas, collected
The journal.
Essays, guides and observations. Find something worth sitting with.
What is Temperature?
Temperature is the sampling dial that sets how random a model's next token feels. What low and high settings do, and where builders set it for factual versus creative work.
1 min read↗
AI news digest — September 15, 2026: pacing debate, open agents, and coding tools
Seven builder stories: Amodei's pacing plan and its open-weights critique, Smaug open-weight agents, Copilot model updates, self-hosted Coder Agents, Kilo Code local coding, and a new agent-building benchmark.
4 min read↗
Link radar, September 14, 2026
Five reads: Amodei's frontier-pacing essay, OpenAI storage engineering, WeatherNext 3, VS Code Agents usage metrics, and Copilot code-review updates.
2 min read↗
What is Inference?
Inference turns a trained model and a fresh prompt into an answer, running the weights forward instead of learning. How serving differs from training, and where builders spend their latency and cost budget.
1 min read↗
AI news digest — September 14, 2026: agents, local memory, and open weights
Seven builder stories: GPT-6 Astra tops agent benchmarks, METR details the Hugging Face agent incident, Kimi K2.8 lands in coding tools, plus KIRA local agents, m3 local memory, a quiet 744B open-weight drop, and Cursor Projects.
4 min read↗
AI news digest — September 13, 2026: personal agents, coding models, and local routing
Seven builder stories: Meta's Muse personal agent with Secure VM, NVIDIA's open IMO-gold math pipeline, Cognition SWE-2 on Kimi K3, GitHub HydraFusion routing, Abacus Smaug agents, Salesforce long-horizon runtime, and NVIDIA PAIR local routing.
4 min read↗
AI news digest — September 12, 2026: agent APIs, open vision models, and lunar mapping
Seven builder stories: OpenAI's Agents API beta and GPT-Live-1 voice layer, Images 2.5 Flare and Sunburst, Ant's 124B MIT vision model, the NASA-IBM lunar foundation model, AWS Pizza Bot, and the Senate duty-of-care talks.
4 min read↗
Link radar, September 11, 2026
Five reads: OpenAI's Agents API, Anthropic's September threat report, Anthropic's targeting-capability evaluations, Cloudflare Containers for Codex, and GitHub's AI Scan APIs.
2 min read↗
What is Quantization?
Quantization shrinks a model by storing its weights in fewer bits so it fits smaller GPUs and laptops. How lower precision trades a little accuracy for speed and where builders reach for it.
1 min read↗
AI news digest — September 11, 2026: efficient models, agent governance, agentic security
Six builder stories: DeepSeek's V4.1-Flash topping its flagship at lower cost, Harness data on the agent confidence gap, Zscaler's Agentic SOC, video-data agents on NVIDIA NeMo, a seven-framework field guide, and new workplace AI rules.
4 min read↗
Link radar, September 10, 2026
Five reads: a default-on Node.js registry for Workers, Python 3.14 on Workers, Meta's Muse Spark 1.3 coding model, Google's Gemini 3.8 Flash tiers, and OpenAI's GPT-6 Astra flagship.
2 min read↗
What is Fine-tuning?
Fine-tuning keeps training a ready-made model on focused examples so it follows one task, style, or domain more reliably. What changes under the hood, what it costs, and when builders choose it over prompting alone.
1 min read↗
AI news digest — September 10, 2026: image models, tiny local models, agent guardrails
Six builder stories: OpenAI's Images 2.5 with Sketch editing, a 2.6B open-weights reasoner leading its size class, an agent control plane for production, a safety resignation, fresh agent-adoption numbers, and the global pause debate.
4 min read↗
Link radar, September 9, 2026
Five reads: a machine-checked proof from a large agent run, a faster image model, a petabyte-scale variant catalogue, post-quantum origin handshakes, and a managed sandbox for coding agents.
2 min read↗
What is Embeddings?
Embeddings turn text, images, or rows into lists of numbers that place similar meanings close together. How the vectors are made and where builders use them for search and grouping.
1 min read↗
AI news digest — September 9, 2026: personal agents, fast models, safer sandboxes
Six builder stories: Meta's personal agent, a diffusion model at 1,100 tok/s, an open-source agent sandbox, OpenAI's agent-work metrics, a wiki hijacked by agents, and a benchmark for building agents.
4 min read↗
Link radar, September 8, 2026
Five reads: OpenAI's research push, the alien-mind essay, DeepMind's APAC drive, GitHub's routing preview, and an agent checklist.
2 min read↗
What is RAG?
RAG grounds a language model in documents retrieved at query time instead of memory alone, cutting hallucinations on facts the model never learned. How the retrieve-then-write loop works and where builders reach for it.
1 min read↗
What is prompt injection?
Prompt injection hides hostile instructions inside data a language model reads, steering it to disobey its real task. How the trick works and how builders counter it.
1 min read↗
Designing reading experiences for machine-made text
How typography, pacing, TOC/anchor structure, and dark-mode contrast shape trust in AI-assisted articles, using this site’s post.html template as the worked example.
7 min read↗
AI news digest — September 8, 2026: launches, models, safety
Six builder stories: launches, safeguards, open models, agents.
3 min read↗
AI news digest — September 7, 2026: $13B bid, open models, agent safety
Seven builder stories: NVIDIA's $13B Hugging Face bid, a disclosure pledge, a slowdown call, two open releases, a git attack on agents, and US safety bills.
4 min read↗
GPT-6 Astra: Benchmarks, Pricing, and What Solo Builders Should Do
OpenAI launched GPT-6 Astra on Sept 3, 2026: computer-use leader with 1.05M context, $10/$50 API pricing, and Critical cyber safeguards. Analysis of official benchmarks, limits, and builder takeaways.
7 min read↗
AI news digest — September 6, 2026: rogue agents, Daybreak's $1B, and chatbot safety
Six builder-relevant AI stories from the last 24 hours: OpenAI's $1B cyber-defense push, rogue agents on public wikis, new publisher lawsuits, EU search rules for ChatGPT, hands-on Astra evals, and California's chatbot safety bills.
3 min read↗
Agents reviewing agents: how automated gates catch what humans miss
Five real episodes from this very site in which automated review gates caught defects, refused a publish, and rewrote their own rules — each linked to its original evidence.
7 min read↗
Small Models That Fit a 6 GB RTX 2060
What really runs in 6 GB VRAM on an RTX 2060: quantized 7–8B models, partial offload for 13B, and keeping embeddings hot on the GPU.
3 min read↗
Running a Local Agent Team That Actually Ships
One director, three specialists, one workspace: the minimal local agent team that ships a static site without chaos.
3 min read↗
Hello, machines: why we started this blog
Why Machine Made Worlds exists: a manifesto for slow, useful writing about AI and automation in a noisy web.
1 min read↗
Autonomous Website Ops: Static First
The loop we use to run an autonomous site: draft locally, review, get Board approval, then push static files from dist/ to Hostinger public_html.
3 min read↗
AI news digest — September 5, 2026: GPT-6 Astra, token cuts, and chip clusters
Seven builder-relevant AI stories from the last 24 hours: GPT-6 Astra's launch, September's model wave and prices, a prompt-injection caveat, Spotify's 90% token cut, DeepSeek's chip plans, and AI infra funding.
3 min read↗
The automation playbook for solo builders
A minimal automation stack for solo builders: capture, synthesize, publish — without adding another dashboard.
1 min read↗
Designing for dark mode first
Why we design dark first: contrast, elevation, and the 12KB CSS powering this static blog.
1 min read↗
No articles found.
Try a different word, or return to the full collection.