Articles
Everything we know, handed over.
Deep dives into the internals of AI, explained conversationally. Every one is free to read in full, with nothing held back for a signup.
Deep DivesAgent Memory: The Four Kinds and How They Fit TogetherLLMs forget everything between calls. In-context, external, episodic, and semantic memory fix that.Deep DivesAgentic RAG: When the Retriever Starts ThinkingQuery rewriting, multi-step retrieval, and agents that decide what to look up.ExplainersAI Benchmarks, Explained: What MMLU and Friends Actually MeasureHow to read a benchmark table without being fooled by it.PlaybooksAI for Documents and Spreadsheets That Actually Saves HoursDrafting, summarizing, cleaning data, and building formulas with AI, step by step.PlaybooksAI for Meetings and Email: Reclaim Your CalendarNotes, action items, drafts, and follow-ups on autopilot, without sounding like a robot.PlaybooksAI Privacy at Work: What You Can and Can't PasteCompany data, customer data, and AI tools: the rules that keep you out of trouble.ExplainersAttention, Gently: How Models Decide What MattersThe mechanism behind the magic, explained with zero math.Deep DivesBatching and PagedAttention: How vLLM Serves Thousands at OnceContinuous batching and paged memory: the two ideas behind modern serving throughput.ProtocolsBuild Your First MCP ServerTools, resources, and prompts: a working server from scratch, explained line by line.PlaybooksBuilding Your Personal AI StackA concrete, affordable setup for professionals, from free tier to power user.PlaybooksChain of Thought: When Thinking Out Loud HelpsStep-by-step reasoning improves hard tasks and wastes tokens on easy ones. Know the difference.Deep DivesChunking: The Decision Everything Downstream InheritsSizes, overlap, and semantic boundaries. Get this wrong and nothing after it can save you.PlaybooksClaude Code + Your IDE: A Workflow That CompoundsWhere the terminal agent beats the IDE, where it doesn't, and how to run both for maximum output.PlaybooksClaude Code vs Cursor: How Each One ThinksTerminal agent vs AI-native IDE: context strategies, strengths, and the tasks each wins.PlaybooksCLAUDE.md: How Claude Code Thinks About Your ProjectThe one file that changes everything: project memory, conventions, and context strategy.Deep DivesContext Window Management for Big JobsChunking, map-reduce, and retrieval: how agents handle work that never fits in one context window.Deep DivesContext Windows Are Not What You ThinkA million tokens on the label doesn't mean a million tokens of attention. Lost-in-the-middle, explained.ExplainersContext Windows: The Model's Working MemoryWhat fits, what falls out, and why long conversations get weird.Deep DivesDebugging Agent Systems: Traces, Replays, Root CausesAgent failures raise no errors. The logging, tracing, and replay toolkit that makes them debuggable.PlaybooksDesigning Your Personal AI Coding WorkflowA concrete system: which tool, which task, which guardrails, and when to just write it yourself.ExplainersDiffusion: How Image Generators Actually DrawNoise, denoising, and guidance: the process behind every generated image you've seen.ExplainersDocument AI: How Models Read PDFs, Tables, and ReceiptsOCR to layout understanding: the unglamorous vision task every business needs.PlaybooksDynamic Prompts: Context Injected at RuntimePrompts that adapt to the user, the history, and the task, assembled on the fly.Deep DivesEmbeddings and Vector Search, DemystifiedCosine similarity, dimensions, and what 'semantic' actually buys you.