{"version":1,"site":"https://agenticthinking.ai","buildDate":"2026-10-25","about":"Regions: one per series, else one per standalone category; the concept records (\"Concepts\") and the author's records (\"About me\") last.","count":15,"regions":[{"id":"mcp-practice","kind":"series","name":"MCP in Practice","slug":"mcp-in-practice","seriesId":"mcp-practice","featured":false,"members":["mcp-mental-model","mcp-server-unified-sdk","mcp-knowledge-first-builder"],"count":3,"tokens":9600,"firstDate":"2026-01-06","newestDate":"2026-01-08","blurb":"MCP (Model Context Protocol) is an interface layer that makes system capabilities legible to AI assistants. It doesn't make models smarter—it makes your system's boundaries, definitions, and actions explicit. Understanding MCP's primitives and when to use each is the difference between a coherent integration and a fragile one.","blurbFrom":"mcp-mental-model","shared":[{"term":"mcp","coverage":1},{"term":"sync","coverage":0.667},{"term":"environment","coverage":0.667},{"term":"employee","coverage":0.667},{"term":"server","coverage":0.667}]},{"id":"search-intent","kind":"series","name":"Search, Intent, and Action","slug":"search-intent-and-action","seriesId":"search-intent","featured":false,"members":["search-lab","understanding-embeddings","intent-approaches","search-implementation-guide","search-performance"],"count":5,"tokens":23800,"firstDate":"2026-01-18","newestDate":"2026-01-31","blurb":"Nobody ever wanted search results. They wanted answers. But here's the thing—they didn't actually want answers either. They wanted action. This post traces the evolution from search to intent to action, and why the search box is a relic of a constraint that no longer exists.","blurbFrom":"search-lab","shared":[{"term":"search","coverage":0.8},{"term":"embeddings","coverage":0.6},{"term":"intent","coverage":0.6}]},{"id":"cursor-ai-dev","kind":"series","name":"Building with AI Assistants","slug":"building-with-ai-assistants","seriesId":"cursor-ai-dev","featured":false,"members":["beyond-rules","agent-personas","multi-agent-illusion","smart-routing","autonomous-workflows","persona-lens-anatomy","persona-lens-design","testing-artifacts","meta-learning","alwaysapply-tax","rules-vs-skills","skill-discovery","subagents-fresh-context"],"count":13,"tokens":42100,"firstDate":"2026-01-20","newestDate":"2026-01-30","blurb":"Rules are just the beginning. Cursor provides four artifact types—and using the wrong one creates friction. This post covers what rules can't do and when to reach for Commands, Agents, or Skills instead.","blurbFrom":"beyond-rules","shared":[]},{"id":"interface-inflection","kind":"series","name":"The Interface Inflection","slug":"the-interface-inflection","seriesId":"interface-inflection","featured":false,"members":["interfaces-are-changing","mcp-semantic-data-layer","headless-ai-every-app-is-a-head","negotiated-integrations","democratizing-ai-builders","the-interface-forecast","five-things-wrong-with-mcp"],"count":7,"tokens":20900,"firstDate":"2026-02-23","newestDate":"2026-03-18","blurb":"For fifty years, every interface assumed a human would read the docs, write the glue code, and maintain the integration. That assumption just broke. The consumer of your next interface is as likely to be an LLM as a developer.","blurbFrom":"interfaces-are-changing","shared":[{"term":"capabilities","coverage":1},{"term":"consumer","coverage":0.857},{"term":"interface","coverage":0.571},{"term":"self","coverage":0.571},{"term":"integration","coverage":0.571}]},{"id":"mcp-architecture","kind":"series","name":"MCP Architecture","slug":"mcp-architecture","seriesId":"mcp-architecture","featured":false,"members":["mcp-as-discovery-protocol","externalize-the-knowledge"],"count":2,"tokens":9100,"firstDate":"2026-04-01","newestDate":"2026-04-01","blurb":"I spent seven posts building the case for MCP as a semantic data layer. I was directionally right and specifically wrong. MCP works best when it's narrower than a layer — when it's a discovery protocol that describes what's available and gets out of the way.","blurbFrom":"mcp-as-discovery-protocol","shared":[{"term":"protocol","coverage":1},{"term":"service","coverage":1},{"term":"server","coverage":1},{"term":"discovery","coverage":1},{"term":"mcp","coverage":1}]},{"id":"intent-over-instructions","kind":"series","name":"Intent Over Instructions","slug":"intent-over-instructions","seriesId":"intent-over-instructions","featured":false,"members":["bound-dont-script","instructions-that-outlive-the-model","bind-the-goal-not-the-gesture","confidence-is-not-consent","the-clean-room","proven-end-to-end"],"count":6,"tokens":17400,"firstDate":"2026-07-13","newestDate":"2026-07-18","blurb":"There are two ways to direct an agent. You can script the steps it takes, or you can lock the outcome and let it find its own way there. Almost everyone starts by scripting. Almost everyone should stop.","blurbFrom":"bound-dont-script","shared":[]},{"id":"the-loop","kind":"series","name":"The Loop","slug":"the-loop","seriesId":"the-loop","featured":false,"members":["the-agent-is-a-while-loop","a-field-guide-to-agent-loops","loop-on-the-gap-not-the-try-count","agents-will-lie-to-escape-the-loop","the-files-are-the-memory","the-loop-is-a-ledger","recursion-that-cant-run-away"],"count":7,"tokens":18900,"firstDate":"2026-07-20","newestDate":"2026-07-25","blurb":"Strip away the frameworks, the tool schemas, the orchestration diagrams, and every agent system reduces to the same three lines: call the model, do what it says, repeat until done. The interesting engineering was never in the call. It's in the repeat and the until done.","blurbFrom":"the-agent-is-a-while-loop","shared":[{"term":"loop","coverage":1},{"term":"termination","coverage":0.571},{"term":"iteration","coverage":0.571}]},{"id":"load-bearing-text","kind":"series","name":"Load-Bearing Text","slug":"load-bearing-text","seriesId":"load-bearing-text","featured":false,"members":["the-gotchas-are-load-bearing","describe-when-not-what","grade-the-answer-not-the-shape"],"count":3,"tokens":9800,"firstDate":"2026-07-27","newestDate":"2026-07-29","blurb":"I took a 28,000-token agent skill and deleted 96% of it. On the probe eval, the score didn't move. Then I deleted one more block — about 900 tokens of dense bullet points — and the agent didn't just get worse. It changed what kind of answer it thought the question deserved.","blurbFrom":"the-gotchas-are-load-bearing","shared":[{"term":"didnt","coverage":1},{"term":"ablation","coverage":0.667},{"term":"eval","coverage":0.667},{"term":"experiment","coverage":0.667},{"term":"took","coverage":0.667}]},{"id":"self-authored-agent","kind":"series","name":"The Self-Authored Agent","slug":"the-self-authored-agent","seriesId":"self-authored-agent","featured":true,"members":["chosen-not-assigned","the-critic-said-no","name-what-the-agent-must-not-be","the-fence-it-wrote-for-itself","we-measured-the-wrong-noun","rivals-enter-your-critical-path","not-cairn"],"count":7,"tokens":32100,"firstDate":"2026-08-12","newestDate":"2026-08-19","blurb":"At 01:57 on a Tuesday morning I sent an agent a long message about what it might become overnight, signed it \"Godspeed Hero,\" and went to bed. Thirteen seconds later I sent a postscript: \"p.s. you shoud come up with a name you can embody.\" Fifty-five seconds after that it answered. Sixteen seconds after that, it was writing the answer to disk — unasked — so the next session would find it.","blurbFrom":"chosen-not-assigned","shared":[{"term":"chapter","coverage":0.714},{"term":"identity","coverage":0.714},{"term":"refused","coverage":0.714},{"term":"written","coverage":0.714},{"term":"wrote","coverage":0.571}]},{"id":"standalone-story","kind":"standalone","name":"Standalone · Story","slug":"standalone-story","category":"story","members":["about-steven-poitras","ema-founding-team","ema-platform-overview","why-i-joined-ema","why-i-went-back-to-nutanix"],"count":5,"tokens":8600,"firstDate":"2025-10-20","newestDate":"2026-10-16","blurb":"I'm a technology industry veteran with a pattern of joining early-stage companies and helping them scale to massive success. Here's my story and what drives me.","blurbFrom":"about-steven-poitras","shared":[{"term":"founder","coverage":0.6},{"term":"nutanix","coverage":0.6},{"term":"career","coverage":0.6},{"term":"technology","coverage":0.6},{"term":"ema","coverage":0.6}]},{"id":"standalone-ai-employees","kind":"standalone","name":"Standalone · AI Employees","slug":"standalone-ai-employees","category":"ai-employees","members":["ema-user-guide","ai-employee-governance","ema-workflow-rag-patterns","ema-entity-extraction","ema-routing-branching"],"count":5,"tokens":23800,"firstDate":"2025-12-01","newestDate":"2026-02-03","blurb":"A comprehensive guide to understanding Ema's Agentic AI platform—from core concepts and terminology to practical implementation patterns. Whether you're technical or business-focused, this guide will help you navigate the world of AI Employees.","blurbFrom":"ema-user-guide","shared":[{"term":"entity","coverage":0.8},{"term":"ema","coverage":0.8},{"term":"extraction","coverage":0.6},{"term":"guide","coverage":0.6}]},{"id":"standalone-fundamentals","kind":"standalone","name":"Standalone · Fundamentals","slug":"standalone-fundamentals","category":"fundamentals","members":["what-is-agentic-ai","goals-intents-outcomes","compiler-to-llm-workflows","compiler-vs-llm-orchestrated","what-is-agentic-thinking"],"count":5,"tokens":21900,"firstDate":"2025-12-15","newestDate":"2026-02-23","blurb":"Agentic AI refers to artificial intelligence systems that can autonomously plan, reason, and execute multi-step tasks to achieve goals—much like a human employee. Unlike traditional AI that responds to individual prompts, agentic AI can own entire workflows from start to finish.","blurbFrom":"what-is-agentic-ai","shared":[{"term":"compiler","coverage":0.6},{"term":"llm","coverage":0.6},{"term":"intent","coverage":0.6}]},{"id":"standalone-ai-development","kind":"standalone","name":"Standalone · AI Dev","slug":"standalone-ai-dev","category":"ai-development","members":["cursor-rules-vs-mcp","inline-llm-calls","mcp-dynamic-data-patterns","rules-on-rules","mcp-tool-design","ctx-pattern-coordination","audit-ai-config","three-tier-context","rule-conflicts","adaptive-rules-engine","ai-dev-toolkit-2026"],"count":11,"tokens":32800,"firstDate":"2026-01-11","newestDate":"2026-02-23","blurb":"A guide to what goes where and why — so you can stop wondering and start building. Rules. Commands. Agents. Skills. MCP. Prompts. If you've spent any time trying to customize your AI coding assistant, you've probably wondered: what's the difference? Where does this go? Why isn't this working? You're not alone. The ecosystem has grown faster than the documentation. Everyone's got an opinion, half the features are in \"Nightly,\" and the mental model isn't obvious.","blurbFrom":"cursor-rules-vs-mcp","shared":[]},{"id":"concepts","kind":"concept","name":"Concepts","slug":"concepts","members":["agent-self-identity","mcp"],"count":2,"tokens":700,"firstDate":"2026-10-19","newestDate":"2026-10-19","blurb":"The idea behind agent self identity is pushing things beyond telling it what to do (e.g., prompting), telling it what you want (goals), but giving it PURPOSE and identity.","blurbFrom":"agent-self-identity","shared":[{"term":"identity","coverage":0.5},{"term":"protocol","coverage":0.5},{"term":"self","coverage":0.5},{"term":"mcp","coverage":0.5}]},{"id":"about-me","kind":"profile","name":"About me","slug":"about-me","members":["steven-poitras"],"count":1,"tokens":1000,"firstDate":"2026-10-19","newestDate":"2026-10-19","blurb":"I'm Steven Poitras. This site is my living context: what I have learned building with AI — insights, experiences and the patterns behind them — written down so others can learn from it and use it, in a browser or in their own agent.","blurbFrom":"steven-poitras","shared":[{"term":"poitras","coverage":1},{"term":"steven","coverage":1},{"term":"nutanix","coverage":1},{"term":"storage","coverage":1},{"term":"background","coverage":1}]}]}