Tips / Practitioner Tips
Practitioner Tips
Short notes on methods AI practitioners have shared — what it is, when to use it, how to try it — with links to the originals.
- Gather your skills and AGENTS.md in one repo and point every other repo to it — one session can then work with the whole system in view The Home Repo Pattern
- Treat video generation as next-frame prediction and judge it on physics and geometry, and you measure how well a model understands the world, not whether it suits someone's creative taste You Can't Tell What's Real Anymore. Here's What Comes Next - Cristóbal Valenzuela, Runway
- Don't settle for a decision model's default 0.50 threshold — tune it on your own labelled data and the same model catches failures it used to miss How To Use Decision Models for Evals
- Plug your own scenarios straight into several models and run them — you see the real differences in your app that benchmarks cannot tell you The Model Swap Workshop: GPT-5.5 vs Kimi vs DeepSeek vs Mistral — Pamela Fox, Microsoft
- Strip out the coding aids in your harness that the model has already learned, and invest in cloud runtime, permission scope, and cost per task — the score stays the same but the cost changes Smarter Models Killed the Harness. Here's What's Left — Jediah Katz, SpaceXAI
- Draw the process as a DAG and turn each node's check into an evaluation — you can see which steps to hand to an agent and which steps humans still hold How to Know When an AI Agent Can Work Unsupervised — Evan Kormos, Coinbase
- Before you build agents, lay down registration, a gateway, memory, tracking and PII blocking as one set — so every agent passes through the same control path Paving the Road to Production: Governed Enterprise AI Agents — Harish Gaggar, Credit Karma
- Before you build, name one thing that is new and yours and put it on the roadmap — and avoid spending two years on the same output as everyone else I Spent Two Years Building the Median — Hariprasad Sudharshan
- Leave control flow, evidence and outbound actions to deterministic code and confine the model to typed decision points — ambiguous requests no longer break the workflow, and compliance can sign off on it Reliable, Smart Workflows for Finance, Wealth and Insurance — Karthic Rao, Perpetual
- Define agents as YAML workflows, not free-running loops — per-step permissions show up in code review and runs become predictable Agents in Orbit: Workflows, Not Autopilot — Eduardo Simon & Javier Casal, Flywire
- Replace static evals with an eval flywheel driven by customer signals — turn post-launch hope into data Why 80% Reliability Isn't Good Enough — Felipe Blanes, Amazon AGI Lab
- Push repetitive work such as summarizing and classification down to Haiku 5.5 — at about 21 cents per task, it cuts cost to roughly a quarter of Sonnet's AI News: Haiku-5.5, Visual ChatGPT, Grok's New Trick, and more!
- Don't just roll out tools — clear two full weeks, half workshops and half real work, and even non-engineers come back with tools people use How a Remote Company Builds AI Fluency — Em Shreve, Automattic
- Let the LLM only decide what to do next and hand execution to a deterministic harness — cluster restarts then run the same way every time Brains vs Hands: How to Run AI Agents Safely in Production — Viren Baraiya
- Break the inference engine down by workload, scheduler and cache, and you can pinpoint the causes of latency and throughput problems from the dashboard alone What Is an Inference Engine, Anyway? — Charles Frye, Modal
- Strip every credential out of the agent sandbox and control tools through a single MCP gateway: if something goes wrong, the damage stays inside that one task How Many Credentials Should Your AI Agent Have? Zero. — Jim Clark, Docker
- Attach scores and Topics labels to your traces, and you can pull unknown failures out of accumulated logs and feed them back into evaluation datasets and PRs Advanced Workshop: Mastering AI Observability — Doug Guthrie, Braintrust
- Implement a spec's tickets in parallel with subagents, and hand off opening a session per ticket and tracking order to a single orchestrator The /implement-spec Skill
- Compare an always-on agent that picks its own tools without model selection against the new half-price models, and rework the model-and-cost mix in your workflow AI News: Dots, GPT-6.1 Sol, Sonnet 5.5, Gemini 4, and everything you need to know