No-Code Agentist

Coding tools changed what’s possible for engineers. We took that same model and asked what could give more Stripes—sales, finance, ops—that same boost. Here’s how we built Stripe’s Knowledge AI Platform. https://t.co/2POTmqjqyD

Every agent workflow is a bet on who proposes, who executes, who checks, and in what order. We looked at 5 systems making that bet differently, from fully hand-built pipelines to workflows discovered by search (ADAS, AFlow). Guess who won ? https://t.co/6ESnZNMIsl

from Stanley - how'd he do?? Eric, went through your last 25 posts against your 44K follower base. Here's the honest breakdown. What's working: Hermes content is carrying your entire account right now. Your top 3 posts by engagement are all Hermes: "Hermes is INSANE as an operating layer" (398 likes, 57K impressions), the "steal this" prompt post (243 likes, 36K impressions), and the OpenClaw vs Hermes debate (120 likes, 19K impressions). When you show your real workflow inside a tool you genuinely use, people pay attention. The pattern is specific: your audience doesn't want to hear that AI is powerful. They want to see exactly how you're using it, what your setup looks like, and what they can copy. The "steal this" prompt post worked because someone could literally paste it into their agent and get value in 60 seconds. That's the format. The OpenClaw vs Hermes post (120 likes, 28 replies) is your highest reply count in the set. Comparison takes with a genuine opinion generate conversation. You said something the Hermes and OpenClaw communities both had a reaction to. That friction is engagement fuel. Your X Article (188 likes, 114K impressions) is your widest reach piece. But the 0.16% like-to-impression ratio tells you the algorithm pushed it far but it didn't convert deeply. Long-form articles on X get distribution but often get saved, not liked. That's still valuable if you're optimizing for brand exposure over vanity metrics. What's not working: 1. Your product launches are getting crushed by your organic content. The ClickFlow MCP announcement (44 likes, 24K impressions) is decent reach, but compare it to Hermes content hitting 57K impressions and 398 likes. Same account, same week, 9x the engagement. The Skills Dojo culture post (10 likes, 956 impressions) barely moved. And the bare "https://t.co/D5WChoT5i7" link? 1 like. 147 impressions. On a 44K-follower account. The algorithm killed it on sight. The problem isn't the products. It's the framing. Your Hermes posts work because they show you using a thing. Your product posts read like announcements. "Introducing ClickFlow for agents" is a press release. "I gave my agent access to Skills Dojo and now it installs its own skills mid-task" is a story. Same product in both cases, but one got 23 likes and the other would have gotten 200+ if you'd led with the workflow, not the product. 2. The hiring post underperformed its potential. "If you're AI-pilled and you're working at an agency, we're building the marketing services as software firm" got 21 likes and 3,392 impressions. That's 7.6% reach on your base. The content was solid, the requirements were clear. But it opened with a conditional ("if you're AI-pilled") that let most people self-select out before reading. A hiring post that opens with the vision ("We're turning Single Grain into a firm where one operator runs an entire client account end to end") would pull more people in before filtering. 3. List posts are underperforming your narrative posts. "14 things that will feel normal inside agencies" got 46 likes but only 3,232 impressions. That's your third-lowest reach in the set for a substantive post. Meanwhile, your narrative posts (the Hermes workflow, the OpenClaw debate, the "one session produced an entire content pipeline") are 5-17x higher reach. The algorithm is telling you something: your audience responds to "here's what I'm doing" more than "here are 14 predictions." Lists feel like content. Stories feel like access. 4. The "just write content" post (12 likes, 2,509 impressions) is your most generic-sounding post. It reads like something any marketing account could post. No personal story, no specific numbers, no "here's what happened when we did this at Single Grain." When you strip away the personal experience, you blend into the timeline noise. Your voice: Operator who shows his actual screen. Your best content doesn't teach theory, it shows process. You're at your strongest when you say "here's exactly what I did, here's the result, here's how you'd do it." The Hermes operating layer post, the content pipeline session, the "steal this" prompt. Those all follow the same DNA: real tool, real workflow, real output, real takeaway. When you drift into announcement mode or prediction mode, the voice flattens. You become interchangeable with 500 other AI-agency founders posting about the future of services. But nobody else is showing their exact Hermes setup with profiles split by thinking/execution/workflow. That specificity is your edge. What to fix: 1. Launch every product through a workflow story, not an announcement. Instead of "Introducing ClickFlow for agents," show yourself using ClickFlow in real time: "My agent just found 14 pages losing traffic, rewrote 3 of them, and dropped them as WordPress drafts while I was on a flight. Here's exactly what happened." Let the product be the answer to the story, not the headline. 2. Stop posting bare links. "https://t.co/D5WChoT5i7" and "https://t.co/udeHNm2Sg7" are algorithmically invisible and make your timeline look like a landing page. Every link needs to be earned. Setup → insight → proof → then the link as the payoff. 3. Lean harder into comparison takes. The OpenClaw vs Hermes post got 28 replies, your highest by far. Your audience wants your honest opinion on tools because you actually use them. "I used Cursor, Claude Code, and Windsurf for a week each. Here's which one I kept and why." "I tested 3 AI SEO tools against ClickFlow on the same site. Here's the data." Comparisons create debate. Debate creates replies. Replies feed the algorithm. 4. Your list posts need a structural rethink. The "14 things" post had genuinely smart insights buried in a format the algorithm apparently deprioritized (3,232 impressions). Take the single most provocative point from that list ("Your agency's GitHub repo matters more than your case study PDF") and make it its own post. One bold claim with a story behind it will outperform 14 claims in a list every time. 5. The podcast is invisible on your timeline. You host Marketing School, one of the most well-known marketing podcasts. In 25 posts, the only podcast mention is the HeyGen experiment (6 likes). That's a massive missed opportunity. Every episode has a moment that could be its own post. "My co-host said something on Marketing School this week that I disagree with:" → instant engagement. The bottom line: You're sitting on a 44K-follower account that performs like a 15K account on most posts, then spikes to 150K+ when you hit the right format. The gap isn't talent or credibility. It's that your best content (real workflows, real tools, real opinions) is mixed with product announcements and generic takes that pull your average down. If every post had the energy of the "Hermes is INSANE" post, this account grows fast. One question: you've built Single Grain into an AI-native agency, you're building SingleBrain, ClickFlow, and Skills Dojo simultaneously. Which one of those bets are you most nervous about, and what would have to be true for it to fail? That answer is the post that would outperform everything on your timeline right now, because your audience is watching someone juggle four things at once and nobody's asking the hard question out loud yet.