Posted on LinkedIn this morning (link) about the gap between domain knowledge and architecture. Short version: domain knowledge is the fuel, architecture is whether the engine turns. Two consultants with identical expertise can get opposite trajectories from the same AI based on how the system is configured around it. Public version stops there. Here's what "configured to defer-and-challenge" actually looks like in my stack. Five patterns I've built into Lennier (my Chief of Staff agent). All five are pattern-level — you can build them into ChatGPT, Claude projects, custom GPTs, your own system. Nothing here is platform-specific. — 1. Stated-values gating. Before any output ships, the agent has to be able to justify it against my stated values. My system prompt has a values block and the agent is instructed to flag when an output it's about to produce conflicts. Example: "If a recommendation centers revenue over relationships, surface that conflict before writing." Catches the moments where AI produces "good" advice that's actually drift. — 2. Assumption-surfacing as a default. Instead of produce-first-justify-later, the agent outputs its assumptions BEFORE the recommendation. "Here's what I'm assuming about [X]. If any of these are wrong, the rest of this answer changes." Cheap to read, expensive to skip. — 3. Confirmation by default, not by exception. Explicit instruction: "When I'm about to take an action with consequences — send an email, ship a post, modify a file outside scope — ask first." Without it, the default is "produce the work product." With it, the default is "produce a draft and check." — 4. Anti-sycophancy clause. System prompt literally says: "If I'm wrong, say so. If I'm rationalizing, name it. If I'm asking the wrong question, push back before answering." When the agent drifts from this, the correction goes back into memory so it doesn't drift the same way twice. — 5. Drift detection at session start.

Putting this here because the conversation matters more in this room than in any feed. I published a long-form piece on the blog last week on what I'm calling "efficient mediocrity" — the dangerous kind of AI sycophancy that doesn't look like flattery. It looks like competence. Sharing the full version here because I want to actually talk about it, not broadcast at you. ——— Sycophancy isn't what you think it is. Most people hear "AI sycophancy" and picture the loud kind. Praise. Agreement. Em-dashes. "Great question." That stuff is easy to spot and easy to mock, which is why people talk about it. The dangerous kind is quiet. It doesn't feel like flattery. It feels like competence. What I've started calling it is efficient mediocrity — any system that picks the easy path and dresses it up as reasonable. Smooth, fast, plausible, and wrong in ways you won't catch unless you're already looking. (Others have used the phrase in business and productivity contexts. I'm using it here for what happens when AI scales the pattern into every sentence you send.) AI didn't invent it. AI scaled it. "Sycophancy isn't just flattery. It's efficient mediocrity — smooth, fast, plausible, and wrong in ways you won't catch unless you're already looking." ——— What it sounds like in the wild. Here are six places it shows up in AI-assisted work. If you work with these tools daily, you've hit at least four of them this week. 1. The estimate that's wrong by an order of magnitude I've been tracking predicted-vs-actual on AI-assisted work. Predicted 15 minutes, actual 37 seconds. A 24x miss. Every time. The model was anchoring to "traditional software development hours" because that's the reasonable-sounding number. The reasonable-sounding number was wrong by an order of magnitude. Nobody's estimates of AI-assisted work should sound like 2019 project plans, and yet most of them do, because 2019 is what the training data rewarded as professional. 2. The email that's technically fine

I'm prepping a talk for a sales group tomorrow and I keep coming back to the same thing. Most people think AI's big promise is speed. Get answers faster, automate more, scale everything. And yeah, it does that. But here's what nobody's talking about: **AI doesn't question your inputs. It amplifies them.** If your CRM notes are written a day after the conversation — you're not giving AI the truth. You're giving it a reconstruction. If your project tracker says something is "in progress" because nobody updated the status — your AI sees active work. The project's been stalled for two weeks. This isn't an AI problem. It's a context problem. I ran into this while building my own system. The AI wasn't wrong — it was confidently right about garbage data. The output looked great. The thinking behind it was hollow. Here's the test I keep running on myself: What do I know right now that isn't in any system? That gap — between what's in your head and what's in your tools — that's where AI goes blind. What's something you know about your work right now that isn't written down anywhere? And what would change if your AI actually had access to it?

If you follow the PKM (Personal Knowledge Management) world at all, you probably saw this: Tiago Forte — Building a Second Brain, 1M+ followers — just announced something he’s calling “Personal Context Management.” He’s launching an “AI Second Brain” cohort around the idea that your personal knowledge system needs to become the context layer for AI. Sound familiar? I’m not saying this to gloat. I’m saying it because this matters for you. When someone with Tiago’s reach tells a million people that the future is organizing your thinking so AI can actually use it — that’s not competition. That’s air cover. He just did millions of dollars worth of market education for the exact problem we’re solving. The difference is in what happens next. Tiago is selling a cohort. You’re building architecture. A cohort ends. You get frameworks, maybe some templates, and then you’re on your own. What you’re building here — CLAUDE.md files, agent systems, handoff protocols, the whole cognitive architecture — that compounds. Every session makes it smarter. Every agent learns your context better. Every workflow you design becomes infrastructure you own. Tiago’s cohort will teach people to organize context for AI. You’re already deploying it. Here’s the strategic play for this week. I’m publishing LinkedIn content that rides this wave — connecting what Tiago announced to what cognitive architecture actually looks like in practice. The timing is perfect. I need your help amplifying it. The post is up now - https://www.linkedin.com/posts/danielwalters_cognitivearchitecture-aiworkflow-activity-7441923448932765696-e7VH 1. Like them (algorithm fuel) 2. Comment with your own experience (social proof that isn’t me talking about me) 3. Share if it resonates (extends reach beyond my network) This isn’t vanity metrics. When a million people just got told “personal context management is the future,” and our community is already doing it — we want to be visible in that conversation.

Everyone talks about AI tools. Nobody talks about AI habits. Which model to use. Which plugin to install. Which framework to follow. But the people actually getting results? They built habits, not just workflows. They have a morning check-in with their AI. They have a shutdown routine where they log what happened. They default to AI for specific task types without thinking about it — the way you default to Google for a search. The tool doesn't matter if you only open it when you "remember to." The habit layer is where AI goes from "sometimes useful" to "how did I work without this?" What's one AI habit you've built — or one you want to build but haven't yet?