# Calvin French-Owen Published articles for Calvin French-Owen. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Small Models Have Arrived DevFeed: [Small Models Have Arrived]() Original publisher: [Read original article]() Author: Calvin French-Owen Published: 2026-08-26T12:00:00Z Content type: opinion Language: en Sources: [Calvin French-Owen]() Topics: [Artificial Intelligence](), [Inference](), [Users]() Tags: [ai](), [business](), [consumer](), [cost](), [fast](), [models]() ### AI overview The article argues that fast, inexpensive, capable small models have improved enough to make consumer AI applications more viable. It also suggests these models can handle much of the responsive, incremental work that businesses perform, while reducing inference costs. ### Source excerpt The fast/cheap/good-enough models have quietly gotten good. That unlocks consumer AI, and most of the "token spewer" work that actually runs a business. ## Durable Objects are Made for Agents DevFeed: [Durable Objects are Made for Agents]() Original publisher: [Read original article]() Author: Calvin French-Owen Published: 2026-07-22T12:00:00Z Content type: opinion Language: en Sources: [Calvin French-Owen]() Topics: [Cloudflare](), [Serverless](), [SQLite](), [V8](), [event driven](), [WebSocket](), [wasm]() Tags: [cloudflare](), [event-driven](), [serverless](), [sqlite](), [v8](), [websockets]() ### AI overview An opinionated technical article explains why Cloudflare Durable Objects can be a strong primitive for building agents and other stateful products. It describes their serverless V8 execution model, per-object SQLite storage, ID-based request routing, event-driven structure, and potential shortcomings. ### Source excerpt I've spent the last few months building almost exclusively on Cloudflare Durable Objects. They're a near-perfect primitive for building agents--here's where they shine, and where they fall short. ## Scaling Context Management: Comparing Fable 5 and GPT-5.5 DevFeed: [Scaling Context Management: Comparing Fable 5 and GPT-5.5]() Original publisher: [Read original article]() Author: Calvin French-Owen Published: 2026-06-13T12:00:00Z Content type: opinion Language: en Sources: [Calvin French-Owen]() Topics: [long-context](), [codex](), [Fable](), [API](), [Caching](), [GPU]() Tags: [api](), [caching](), [codex](), [context](), [fable](), [gpu](), [long-context]() ### AI overview The article compares how Fable 5 and GPT-5.5 scale context management for difficult, long-running tasks. It describes Codex's server-side compaction approach and discusses its implications for clients and long-running threads. ### Source excerpt Comparing two very different approaches to scaling context between Fable 5 and GPT 5.5. ## Coding Agents in Feb 2026 DevFeed: [Coding Agents in Feb 2026]() Original publisher: [Read original article]() Author: Calvin French-Owen Published: 2026-02-17T12:00:00Z Content type: opinion Language: en Sources: [Calvin French-Owen]() Topics: [AI-assisted coding](), [context window](), [context](), [Claude Code](), [codex](), [Language models](), [cursor]() Tags: [2026](), [agents](), [claude-code](), [codex](), [coding-agents](), [context](), [cursor](), [tool-use]() ### AI overview An opinionated February 2026 snapshot of using coding agents, especially Claude Code and Codex. The author explains how available time, desired autonomy, model strengths, context windows, compaction, and filesystem-based planning influence tool choice and workflow. ### Source excerpt My point-in-time snapshot of using Claude Code and Codex together. Opus shines at context management and tool use, Codex writes fewer bugs. Here's how I use both. ## How I Use Obsidian DevFeed: [How I Use Obsidian]() Original publisher: [Read original article]() Author: Calvin French-Owen Published: 2026-01-07T22:00:00Z Content type: tutorial Language: en Sources: [Calvin French-Owen]() Topics: [Obsidian](), [Local-First](), [Markdown](), [Claude Code]() Tags: [local-first](), [markdown](), [notes](), [obsidian](), [research](), [setup](), [template]() ### AI overview A personal guide to using Obsidian as a local-first Markdown environment for thinking and writing. The author's central principle is that note-taking should elicit thought, with minimal emphasis on organization and automation beyond practical needs. ### Source excerpt Obsidian has become my personal operating system for thinking and writing. Here's the golden rule I follow and the systems I use. ## You Still Need to Think DevFeed: [You Still Need to Think]() Original publisher: [Read original article]() Author: Calvin French-Owen Published: 2025-09-26T15:15:04Z Content type: opinion Language: en Sources: [Calvin French-Owen]() Topics: [AI-assisted coding](), [coding](), [context](), [codex](), [Claude Code](), [cursor](), [Large Language Model]() Tags: [claude-code](), [codex](), [coding-agents](), [context](), [cursor](), [error-handling](), [llms](), [technical]() ### AI overview This opinion argues that increasingly capable, long-running coding agents do not eliminate the need for human thinking; they change where that thinking is spent. Remote, interactive, and IDE-focused tools shift effort among providing context, planning, implementation, verification, and review. The author suggests LLMs are currently strongest at implementation and weakest at supplying organizational context. ### Source excerpt When it comes to coding agents, the product shape fundamentally shifts how you think. ## The Coding Agent Metagame DevFeed: [The Coding Agent Metagame]() Original publisher: [Read original article]() Author: Calvin French-Owen Published: 2025-08-26T15:15:04Z Content type: opinion Language: en Sources: [Calvin French-Owen]() Topics: [Claude Code](), [AI-assisted coding](), [Command-line interface](), [User interface design]() Tags: [claude-code](), [cli](), [coding](), [coding-agents](), [design]() ### AI overview An exploration of why developers find Claude Code engaging, focusing on its hackable workflows, retro gaming aesthetic, lightweight terminal-based interface, and distinction from traditional IDEs. ### Source excerpt Using Claude Code feels more like playing the piano than shuffling tickets in Jira. I get the vague sense that by using the tool differently, I could become a virtuoso. Here's my best attempt to explain why that is. ## Reflections on OpenAI DevFeed: [Reflections on OpenAI]() Original publisher: [Read original article]() Author: Calvin French-Owen Published: 2025-07-15T15:49:56Z Content type: opinion Language: en Sources: [Calvin French-Owen]() Topics: [OpenAI](), [codex](), [Large Language Model]() Tags: [culture](), [llms](), [openai](), [organization](), [reflections](), [slack](), [work]() ### AI overview A former OpenAI employee reflects on the company's culture, rapid growth, organizational challenges, and differing experiences across research, applied, and go-to-market teams. ### Source excerpt There's a lot of smoke and noise around what OpenAI is doing, but not a lot of first-hand accounts of what the culture of working there actually feels like. Here's my view on what it was like to work there. ## Upsides and Downsides DevFeed: [Upsides and Downsides]() Original publisher: [Read original article]() Author: Calvin French-Owen Published: 2025-02-27T00:00:00Z Content type: opinion Language: en Sources: [Calvin French-Owen]() Topics: [Requirements](), [Security](), [audit](), [Latency](), [Artificial Intelligence]() Tags: [audit](), [churn](), [latency](), [requirements](), [risk](), [security](), [startups](), [velocity]() ### AI overview This opinion article presents a framework for understanding how startup priorities change as a company matures. Early-stage startups benefit from pursuing high-upside opportunities and accepting variance, while later-stage companies must reduce downside risks such as security issues, access-control gaps, audit requirements, uptime problems, and cost or performance concerns. The article also argues that this shift helps explain why AI demos often fail to become strong products. ### Source excerpt When to shift mentality from upsides to downsides. And why most AI demos don't make great products. ## How a manager mindset shapes LLM-assisted coding DevFeed: [How a manager mindset shapes LLM-assisted coding]() Original publisher: [Read original article]() Author: Calvin French-Owen Published: 2025-02-11T12:00:00Z Content type: opinion Language: en Sources: [Calvin French-Owen]() Topics: [Vibe coding](), [Large Language Model](), [cursor](), [engineering-leadership](), [coding]() Tags: [ai-coding-assistants](), [cursor](), [engineering-leadership](), [opinion](), [vibe-coding]() ### AI overview The author examines why engineers differ in their views of LLM-generated code. Drawing on experiments with vibe coding and AI coding assistants, the article argues that success depends partly on how engineers augment themselves with LLMs and how much of the work involves defining goals, guardrails, and desired outcomes rather than implementing technical details. ### Source excerpt There's a lot of debate around how much 'good code' LLMs can write for you. Some engineers claim coding with LLMs is amazing, while others think they are trash. Ultimately, it depends on how much of a manager you are.