James Sowers 📦

That plus when I'm personally driving Claude Code I'm way more likely to have it ask me questions or build a plan first or share a much longer prompt/examples. With OpenClaw it's positioned as this "do it all for you" kind of solution so I just ask my lead agent to make shit happen and then I'm mildly annoyed when it comes back not looking amazing after the first try. 😂

My OpenClaw is managed by missioncontrolhq.ai because I didn't have a spare device laying around and didn't want to figure out how to replicate their agents and kanban-style task management UI for myself. It's on Claude, but it uses Sonnet for token cost. Still figuring out how to be more efficient with tokens but I was maxing out every single session when I used Opus.

Here's another tweet that I fed into my claude code Don't waste your braincells trying to decode tweets like this, just ask claude! My specific suggestions from the tweet: --------- Good tweet — a few ideas from it that would actually improve the skill: 1. Add frequency as a leading fatigue indicator We currently detect fatigue reactively (CTR already dropping, CPM already rising). Frequency > 3.5 is a predictive signal — the audience is getting cooked before metrics tank. The Meta Ad Creative sheet likely has frequency data. Adding it to the creative pull and flagging ads approaching 3.5 would catch fatigue 3-5 days earlier. 2. CPA-based bleeder threshold We have CTR-based kill signals (Link CTR < 2.0% after 25+ clicks) but no explicit spend-bleed detector. Adding something like "CPL > 2x target for 3+ consecutive days → immediate pause recommendation" would catch money pits that have fine CTR but garbage conversion. Different signal, catches different problems. 3. Winner pattern analysis section We do hook category analysis already, but the tweet's "analyze what's working across winners" idea could be expanded. After identifying top 5 ads, a new section like "What's Working" that pulls out common traits: hook style, video length, angle, CTA type. Makes the "what to test next" recommendation concrete instead of generic. 4. Budget shift recommendations with specific numbers We rank by efficiency but don't say "shift $X/day from [worst ad set] to [best ad set]." Adding dollar-specific reallocation suggestions makes the recommendation immediately actionable. 5. Daily health check (separate automation, not the weekly) The biggest gap isn't in the weekly — it's that nothing checks between weeklies. A lightweight daily launchd job that pulls yesterday's spend + CPL per ad set, flags anything bleeding, and sends a 3-line Telegram alert. Like a mini /meta-weekly that runs every morning. Separate skill/script, but the tweet's real win is the daily cadence.

Yeah it kind of feels like 1998 again. All the files live on the desktop. Your computer has to stay on all the time for anything to get done. Your constantly stressing about saving in-progress work and it feels like one wrong lightning strike will brick your computer and take your whole life/business down with it. I sync everything with iCloud and Google Drive and I still live in a constant state of fear. Haha

@jamessowers I use scheduled tasks but it runs locally so my computer has to be on

@dennishegstad @vercel Claude Code has me on Sanity.io for the CMS + Astro + Vercel for the marketing site...code all hosted on GitHub. 100% happy so far.

Where would you deploy and host a blog? I can't justify the @vercel usage costs for a simple blog I was testing some AI stuff with. $50+ a month in usage costs to deploy a blog without a CMS or any back-end is not interesting. Not a fan of webflow anymore... Where are you hosting blogs with +1000s of pages?

I'm migrating from a super simple Carrd website and this is very much a side project, so it's far from a finished product, but... hiddenhands-gamma.vercel.app This is my current, Cardd-powered website... hiddenhands.co I basically said, "Use the copy/structure of my existing website as a starting point. Here a four Framer/Webflow templates I like. Let's build it using Option 1 and then we can re-skin it if we need to." For the blog, I'm pretty sure I just shared a URL from the Buffer blog and said, "Make it look like this page structure, but without the featured image in the hero section and without the social sharing widget on the left sidebar." It's been very enjoyable so far. I'm still not 100% in love with not being able to make changes myself, but Claude seems to understand and apply my verbal instructions just fine.

@jamessowers @JamesonCamp Can I see what you built with it?

Just canceled ClickFunnels, Circle, Webflow, and Notion in the same month $600+ month in SaaS. Gone Spent probably $20k+ on these tools over the years. Was loyal to all of them Manus builds a better funnel in 20 minutes than what I spent hours dragging and dropping in CF. Better design. Better copy. Actually understands the offer SaaS isn't dead. Most people will use these tools forever But there's a growing group of us in this corner of the internet quietly canceling everything and rebuilding it with AI in an afternoon Kinda wild to mass cancel tools I used to think were essential...

@binghott @JamesonCamp The stack Claude Code recommended for me was GitHub+ Astro + Sanity.io (CMS) all deployed to Vercel and it has been really nice! I'm a non-technical marketer and Claude Code walked me through step by step.

@JamesonCamp Thanks! Yeah, I'm a little concerned about losing the little bit of blog steam my site has, but I don't think it matters that much. I'm sure Claude will take care of all of it for me and/or hold my hand through it.

@jacob_posel Same

I’ll pay someone for their time to show me how they’re using Manus to make landing pages

@pbteja1998 Looks great! Is better Docs search on the roadmap? Would be hugely beneficial for me.

New MissionControlHQ.ai dashboard looks so cool 👌

100% 😂 He built Mission Control to make his personal OpenClaw instance more usable and more performant. Then he basically productized the operating system and made it so that anyone else can create their own version of Mission Control. Now he's pretty transparent about using Mission Control to grow Mission Control, which is pretty cool. Just loving the Wild West nature of all of this!

@jamessowers My favorite tell-tale sign of an AI generated landing page (I know cause half my page have them too haha)

I don't understand this. Every site I've visited looks the same, none of a sign up or pay link, and there's no "product" in sight. Just visit their live ticker page and click any of the sites that pop up

Yeah I agree. For what it's worth, in my limited test it seemed overeager and suggested a bunch of stuff that was mostly busywork and not directly tied to revenue. Not saying that's intentional, but I can see how this would burn through task credits without actually accomplishing much of anything if you don't put up tight guardrails. I like my MissionControlHq.ai setup 10x better. Still a lot of autonomy, but I feel like I have constant control of what gets worked on, who does it, and how much I'm spending on it. (Not sponsored. Not an affiliate. Have zero financial or personal biases related to sharing that tool.)

@jamessowers Right but even at that rate, pretty much all their users would need to be paying users which seems unlikely

@everestchris6 BOT

My OpenClaw + Arcads bot runs multiple Instagram accounts, each with a different AI persona, all driving traffic to my SaaS. It is already pulling thousands of views every week. Here's how it works: → OpenClaw researches trending topics and writes the scripts → Arcads generates videos with realistic AI actors as the face of each account → Multiple accounts, multiple personas, all posting automatically → Every post ends with a CTA to sign up → Zero manual work. Runs 24/7. This is what automated distribution looks like in 2026 Reply "BOT" and I'll send you the full setup and a guide for free

@JamesonCamp Sounds like if I'm just starting with AI web dev today I should be going Replit over Lovable?

We are 99% of the way to letting generalists and ecom people execute $50k of dev work and $10k of design work in a weekend AI closed the coding gap a while ago. The problem was always design. Your code worked...it just looked like a developer built it. I've been using Manus to design Shopify funnels and it's actually great for that. But the second you try to edit a specific element, you're basically starting over. Replit just cracked this: 1. Tell it what you want 2. It spins up 3 design variations at once 3. Pick the one you like most 4. Actually edit the individual elements without the whole thing breaking Step 4 is the unlock. Nobody cared that AI could generate a landing page. We cared that we couldn't touch it after without blowing it up. Oh and by the way, you can have multiple agents building in parallel while you're editing. So while you're dialing in the design on one page, the next three are already being built. If you're a Figma power user this won't move you. But if you're a small team building funnels, landing pages, or apps and you know the loop of generate, hate the design, regenerate, lose your edits... That loop is over.

@alexgoughcooper How long til we get the how-to video on this workflow? I keep trying and failing with small parts (Ex: text formatting/placement). Feels like I'm the only kid left in the class who hasn't passed the test to move from addition and subtraction to multiplication.

Nano Banana 2 is beyond cracked. There is not a static ad out there that can't be one-shot with Claude + Higgsfield + NB2

What would you estimate the total cost of AI credits/usage was here? Assuming you could be 20-50% more efficient if you picked up the same project again right now, but I think it's helpful for folks to know as a benchmark. Guessing between direct AI costs, your hourly rate, and disruption to your lifestyle this was a low five figure investment.

Finally got to a point where we using our new vibe coded OS for all internal processes / client work. We are ditching: Notion Frame Google Docs (parts of) Slack Claude Skills Zapier Will soon be ditching: Forecasting spreadsheets Meta reporting apps Only took two straight months of vibe coding and terrible sleep

@lkr @helloitsolly Worth it over Cowork? I'm not running a big SaaS company. I run a few small productized services. Cowork had been pretty good so far, but feels like I might be leaving something on the table.

The rush of converting someone to claude code Got @helloitsolly who's next??

"You need to automate yourself out of your job." I agree with like 99% of what you publish but I can't get on board with this take right now, at least not as it's phrased here. Chances are your employment contract has some kind of a clause that basically says anything you create while working for the company or using their resources is owned by them, not you. So if you truly automate yourself out of a job, then that seems like the perfect opportunity for them to keep all of those fancy automations you've built and cut your job or give it to somebody cheaper. Not saying this is bad advice at all but I think it only works in your favor if you automate yourself out of a job in such a way that the company gets even more value from the work that you do. There's still some level of dependency on you to make sure it keeps happening without compromising quality or outcomes. Most people who work at companies big enough to lay people off due to AI should probably understand this thing is an entity focused on competition and survival. Despite how much they might say about you being family or a critical part of their success, they will probably cut the expense of your salary at the first chance they get as long as it doesn't increase their risk profile or cost them money. Not fully disagreeing with your take but I think there's important nuance that people should know. If you're going to truly automate yourself out of a job, you better be saving a copy of all of that intellectual property and it would be best if you're working on some kind of a side hustle or contingency plan before you share the results.

Anthropic just published a 17-page report on AI's impact to jobs. It's worth reading, but here are some of the key points made & my observations: Key points: 1) The gap between what AI can do and what people are doing with it is massive. In Computer & Math jobs, AI could handle 94% of tasks. Actual usage is 33%. 2) This isn't coming for factory workers first. The most exposed workers are older, more educated, and higher-paid, earning 47% more than unexposed workers. AI is aimed squarely at white-collar knowledge work. 3) No mass unemployment yet. Zero systematic increase in unemployment for highly-exposed workers since ChatGPT launched. But... 4) Young workers are already getting squeezed. Job finding rates for 22-25 year olds entering high-exposure jobs dropped ~14%. The labor market isn't contracting from the top. Tt's narrowing from the bottom. 5) 30% of workers have literally zero AI exposure. And 97% of actual AI usage falls within what's already theoretically possible. People aren't pushing boundaries of the technology. 6) Most at-risk roles: Computer Programmers lead at 75% coverage, followed by Customer Service Reps, Data Entry Keyers (67%), and Financial Analysts. 7) Least at-risk roles: Cooks, Motorcycle Mechanics, Lifeguards, Bartenders, Dishwashers, Dressing Room Attendants. Jobs where your hands, your body, or your physical presence is the work. My observations: 1) The gap between theoretical and actual AI usage isn't a technology problem. It's a knowledge and process problem. Most people play with a tool for 30 minutes, don't get the output they expect, and decide the technology is dumb and broken. The person is broken, not the technology. You wouldn't hand an intern a laptop on day one and expect killer output with zero onboarding. But that's exactly what people do with AI. 2) Junior workers are getting hit first because the cost of failure is lowest. Companies will trust AI for scheduling, meeting prep, and basic engineering work before they'll trust it for strategic decisions. No exec wants to be the schmuck who trusted a hallucinating model and bombed their board meeting. So the work that gets automated first is the work that juniors/offshore talent used to do. 3) The calm before the storm is real. I believe double-digit percentages of white-collar workers will end up displaced. The only reason we aren't seeing it yet is because most companies haven't done the hard work. Rethinking org design from first principles, mapping every key process, building the infrastructure to actually run leaner. They can't confidently say "if we let go of 20% of our people today, would the business still run?" But this is a knowledge and process issue, not a technology issue. And it will be figured out. because you'll be competitively disadvantaged if you don't. 4) There's a great irony happening before us. Most humans behave like robots. Follow marching orders. Forget to reflect. Don't improvise. Wait for instructions. Meanwhile, robots are behaving more like humans. Generative. Deep thinking. Context-aware. Making adjustments as needed. 5) You need to automate yourself out of your job. Seriously. If you succeed, you've AI-proofed yourself. - You become the person who orchestrates the work, not the person the work replaces. - And you become the shining star in your company — the person leadership looks to and asks "how do we replicate what you've done across every role?" 6) This sounds scary, but new roles will be created. As with every technology paradigm, job destruction begets job creation. AI is different and has its own shape, but there will absolutely be new roles available to those who stay on the frontier. Now, the question to debate is how many new roles & do they offset lost roles. But I can imagine roles focused on orchestration, judgement, and adoption becoming more and more prevalent: - AI Process Architect - Agent Operations Manager - AI Quality Assurance/Output Auditor - AI Onboarding Specialist - Decision Architect - AI Ethics & Risk Officer - Synthetic Media Producer - Knowledge Curator

@TechWith_Nova CLAUDE

@jackfriks Probably switching from Publer when analytics are available for most of the major platforms. Need my agents to analyze performance and use it to get better over time.

you can now hookup your claude cowork to post to all your social media accounts too btw $14/month @ post-bridge.com/mcp