# Matthew Mead

> Practical AI, from someone who still builds.


## Pages

- [TEDx Proposal](https://matthewtmead.com/tedx/)
- [Speaking](https://matthewtmead.com/speaking/)
- [Insights](https://matthewtmead.com/insights/)
- [Home](https://matthewtmead.com/)
- [About](https://matthewtmead.com/about/)



## Posts

- [Thinking Machines releases Inkling 2 Small, with open weights coming](https://matthewtmead.com/2026/11/07/thinking-machines-releases-inkling-2-small-with-open-weights-coming/)
- [Cloudflare lays out a trust framework for AI agents](https://matthewtmead.com/2026/11/07/cloudflare-lays-out-a-trust-framework-for-ai-agents/)
- [METR: AI is now speeding up AI research itself](https://matthewtmead.com/2026/11/07/metr-ai-is-now-speeding-up-ai-research-itself/)
- [Counterfactual](https://matthewtmead.com/2026/11/06/counterfactual/)
- [AI: Beyond Generation](https://matthewtmead.com/2026/11/04/ai-beyond-generation/)
- [The Human Is the Bottleneck Now. Design for It.](https://matthewtmead.com/2026/11/03/the-human-is-the-bottleneck-now-design-for-it/)
- [Two-Way Door Decision](https://matthewtmead.com/2026/10/30/two-way-door-decision/)
- [Small Language Model](https://matthewtmead.com/2026/10/23/small-language-model/)
- [Writing Code Got Cheap. Verifying It Got Expensive.](https://matthewtmead.com/2026/10/21/writing-code-got-cheap-verifying-it-got-expensive/)
- [Evals](https://matthewtmead.com/2026/10/16/evals/)
- [Tokens and Context Window](https://matthewtmead.com/2026/10/09/tokens-and-context-window/)
- [Learning Before Leverage](https://matthewtmead.com/2026/10/06/learning-before-leverage/)
- [AI Agent](https://matthewtmead.com/2026/10/02/ai-agent/)
- [Agents Are the New Attack Surface](https://matthewtmead.com/2026/09/30/agents-are-the-new-attack-surface/)
- [The AI Sunday Drive](https://matthewtmead.com/2026/09/26/the-ai-sunday-drive/)
- [Prompt Injection](https://matthewtmead.com/2026/09/25/prompt-injection/)
- [Hallucination](https://matthewtmead.com/2026/09/18/hallucination/)
- [AI Amplifiers](https://matthewtmead.com/2026/09/17/ai-amplifiers/)
- [Context Engineering](https://matthewtmead.com/2026/09/11/context-engineering/)
- [The AI Cyber Threat](https://matthewtmead.com/2026/09/10/the-ai-cyber-threat/)
- [Human in the Loop](https://matthewtmead.com/2026/09/04/human-in-the-loop/)
- [Innovation with AI](https://matthewtmead.com/2026/09/03/innovation-with-ai/)
- [Probabilistic vs. Deterministic](https://matthewtmead.com/2026/08/28/probabilistic-vs-deterministic/)
- [Harness](https://matthewtmead.com/2026/08/21/harness/)
- [Claude's Secret Watermark](https://matthewtmead.com/2026/08/11/claudes-secret-watermark/)
- [AI Maturity Changes Success](https://matthewtmead.com/2026/08/03/ai-maturity-changes-success/)
- [The 700 HP Problem](https://matthewtmead.com/2026/07/23/the-700-hp-problem/)
- [I Thought This Was AI](https://matthewtmead.com/2026/07/11/i-thought-this-was-ai/)
- [Demo Mode Is Expensive](https://matthewtmead.com/2026/07/07/demo-mode-is-expensive/)
- [Did the Meaning Survive?](https://matthewtmead.com/2026/06/22/did-the-meaning-survive/)
- [AI vs. Law Professors](https://matthewtmead.com/2026/06/05/ai-vs-law-professors/)
- [When Code Stops Being a Craft](https://matthewtmead.com/2026/05/29/when-code-stops-being-a-craft/)
- [Stop Optimizing AI Too Early](https://matthewtmead.com/2026/05/26/stop-optimizing-ai-too-early/)
- [AI Isn't Evil. It's Worse.](https://matthewtmead.com/2026/05/18/ai-isnt-evil-its-worse/)
- [AI Fails Differently](https://matthewtmead.com/2026/05/11/ai-fails-differently/)
- [Don't Trust One AI](https://matthewtmead.com/2026/04/29/dont-trust-one-ai/)
- [Less Autonomy, More Trust](https://matthewtmead.com/2026/04/22/less-autonomy-more-trust/)
- [13 Tools, 1 PDF, and the Truth](https://matthewtmead.com/2026/04/18/13-tools-1-pdf-and-the-truth/)
- [Your AI Problem Probably Isn't the Model](https://matthewtmead.com/2026/04/14/your-ai-problem-probably-isnt-the-model/)
- [Stop Chasing Better Models](https://matthewtmead.com/2026/04/10/stop-chasing-better-models/)
- [Where NOT to Use AI](https://matthewtmead.com/2026/04/03/where-not-to-use-ai/)
- [The AI Reversal](https://matthewtmead.com/2026/03/27/the-ai-reversal/)
- [AI Guardrails, Tested on Gabbo](https://matthewtmead.com/2026/03/21/ai-guardrails-tested-on-gabbo/)
- [What AI Could Not Solve](https://matthewtmead.com/2026/03/13/what-ai-could-not-solve/)
- [Why AI Projects Fail](https://matthewtmead.com/2026/03/06/why-ai-projects-fail/)
- [AI Cost Estimation](https://matthewtmead.com/2026/02/27/ai-cost-estimation/)
- [Few-Shot Prompting](https://matthewtmead.com/2026/02/14/few-shot-prompting/)
- [The Best AI Is the AI You Don't See](https://matthewtmead.com/2026/02/10/the-best-ai-is-the-ai-you-dont-see/)
- [OpenClaw](https://matthewtmead.com/2026/02/03/openclaw/)
- [Chess and AI](https://matthewtmead.com/2026/01/28/chess-and-ai/)
- [OWASP LLM Vulnerabilities](https://matthewtmead.com/2026/01/22/owasp-llm-vulnerabilities/)
- [The Impact of AI Adoption](https://matthewtmead.com/2026/01/16/the-impact-of-ai-adoption/)
- [My 2026 AI Prediction](https://matthewtmead.com/2026/01/09/my-2026-ai-prediction/)
- [AI Became Normal](https://matthewtmead.com/2025/12/30/ai-became-normal/)
- [The AI Autonomy Study](https://matthewtmead.com/2025/12/26/the-ai-autonomy-study/)
- [AI and OCR](https://matthewtmead.com/2025/12/20/ai-and-ocr/)
- [Confidence Is Not Correctness](https://matthewtmead.com/2025/12/11/confidence-is-not-correctness/)
- [LLM as a Judge](https://matthewtmead.com/2025/12/08/llm-as-a-judge/)
- [Think Training](https://matthewtmead.com/2025/11/19/think-training/)
- [AI Browsers: Comet and Atlas](https://matthewtmead.com/2025/10/28/ai-browsers-comet-and-atlas/)
- [Messy Work Is the Moat](https://matthewtmead.com/2025/09/30/messy-work-is-the-moat/)
- [Jailbreaking LLMs](https://matthewtmead.com/2025/09/16/jailbreaking-llms/)
- [Three Curves Every Leader Should Be Watching](https://matthewtmead.com/2025/07/22/three-curves-every-leader-should-be-watching/)
- [Meet Gabbo](https://matthewtmead.com/2025/06/19/meet-gabbo/)
- [MateoBot](https://matthewtmead.com/2025/06/06/mateobot/)
- [GenAI Use Case Categories](https://matthewtmead.com/2025/05/30/genai-use-case-categories/)
- [The Goldilocks AI Problem](https://matthewtmead.com/2025/05/24/the-goldilocks-ai-problem/)
- [Messy AI](https://matthewtmead.com/2025/05/16/messy-ai/)
- [Agents Are Coming for Your Front Door](https://matthewtmead.com/2025/05/06/agents-are-coming-for-your-front-door/)
- [Manus AI](https://matthewtmead.com/2025/04/18/manus-ai/)
- [Bigger Isn't a Strategy](https://matthewtmead.com/2025/04/08/bigger-isnt-a-strategy/)
- [The AI Whisperer (April Fools 2025)](https://matthewtmead.com/2025/04/01/the-ai-whisperer-april-fools-2025/)
- [AI's Version of Moore's Law](https://matthewtmead.com/2025/03/26/ais-version-of-moores-law/)
- [The Agent Pricing Debate](https://matthewtmead.com/2025/03/20/the-agent-pricing-debate/)
- [Post-Quantum Cryptography Is the New Y2K](https://matthewtmead.com/2025/03/13/post-quantum-cryptography-is-the-new-y2k/)
- [Rereading The 4-Hour Workweek](https://matthewtmead.com/2025/03/07/rereading-the-4-hour-workweek/)
- [Operator](https://matthewtmead.com/2025/02/28/operator/)
- [B2A](https://matthewtmead.com/2025/02/14/b2a/)
- [Diminishing Returns](https://matthewtmead.com/2025/02/10/diminishing-returns/)
- [DeepSeek](https://matthewtmead.com/2025/01/30/deepseek/)
- [AI and Energy](https://matthewtmead.com/2025/01/24/ai-and-energy/)
- [What Can't AI Do?](https://matthewtmead.com/2025/01/15/what-cant-ai-do/)
- [Plaud Pin](https://matthewtmead.com/2025/01/10/plaud-pin/)