March

@theonlyhaitham The problem with amazon is horrible sellers treatment, constantly shutting one or the other down, the worst ux in internal systems, verifications that in order to pass you have to fake shit. On the other hand, meta is fucked up as shit too.

The best decision I made in 2025 was hitting the brakes on Meta and Shopify and going all in on Amazon. We're now doing in one month what we used to do in an entire quarter on Shopify. Amazon already has the traffic. So what happens is you stop fighting for attention and start competing on relevancy and execution. And that shift changes everything about where you focus, because when you're not chasing traffic, you're focused on supply chain and operations, and on making sure you can actually fulfill what the algorithm wants to send you. Sales get more sales, Reviews go up, Rank goes up, Organic kicks in, and the flywheel starts moving. I believe most early CPG brands spread themselves too thin across channels, trying to "build a brand" before they've built a business, and Amazon forces you to get the fundamentals right first.

@hunvreus For sure it’s bs. Most of those guys need therapy session

Talking to smarter folks than me, I'm convinced many of the AI folks in my timeline are full of shit. Nobody is "running 20 agents over night" and building stuff for actual users. Maybe some are building internal tools or disposable software. Maybe. But building software people like using? That doesn't get hacked on day one or blow up after the 3rd user? Nope. I don't even understand what that's supposed to look like. Do you work out a 57 pages document that perfectly describes what you want to build and then summon 14 agents and have them run wild for 6 hours? And what comes out on the other end isn't a broken pile of shit? Nope. Not buying it. PS: it may also be that I have an IQ of 82 and can't figure it out.

@harleyf @kinngh @ShopifyDevs Weeks? It’s two months already

When you tell us something's broken, we fix it. App review times in the Shopify App Store got too slow. Devs were waiting weeks. So we rebuilt the submission pipeline.

@codyschneider YES!

I never want to touch a UI again

@codyschneider I wanted email marketing flows for myself with openclaw or claude code. I did it with claude code. crazy how powerful it turned out. No more Klaviyo. Productizing it now leannoku.com

if you build an email marketing platform that allows you do everything through the API you will make so much money

@iamkevinkrieg leannoku.com

Can somebody vibecode a triplewhale clone already? Way too expensive at scale

@Nugennath @klaviyo If you’re up to using claude code,or other agentic tools like OpenClaw, to run your campaigns, I built the sending layer so agents can send directly through API. Same deliverability at a fraction of the cost. I’m using it myself for marketing emails.

8 figure brands are email and sms on Klaviyo Or sms on another app? Orrrr Omnisend? What is the crack I just hate giving Klaviyo all this money @klaviyo

Your website has a new visitor. It doesn't click, scroll, or feel. It reads everything and decides in milliseconds whether to recommend you. Marketing pages are built for humans. Short, punchy, designed to make you feel something in 5 seconds. AI agents are a completely different reader. They consume thousands of words. They don't need hooks. They need detail. And AI queries tend to be longer and more specific: "I need a tool that pulls all my e-comm and ad data into one place so my AI agent can access it." Your hero section doesn't answer that. So I rebuilt my homepage to serve two audiences from the same URL. Here's what I did. The same homepage URL checks the Accept header. If an agent sends "Accept: text/markdown", it gets a 1,400-token markdown document instead of the HTML page. Full docs: every endpoint, every parameter, response shapes, error formats, pricing, integrations. One request, complete picture. Not every AI agent can set custom headers. So I added fallbacks: /index.md is a direct URL that always returns the markdown version. Any agent can just fetch it. /llms.txt is a shorter, curated overview following the llms.txt convention that's gaining traction. Think of it as a README for AI. robots.txt includes Cloudflare's new Content-Signal spec, which explicitly tells crawlers: yes, you can use this content for AI training, search, and grounding. Four ways an agent can get structured content from the site. The markdown reads like documentation, not copywriting. Structured facts an agent can compare and cite. Took me an afternoon. Probably should have done it months ago.

@MatthewBerman @agentmail Building an api layer for ecom stores that AI agents can actually query. No dashboards, just clean data from meta, google, shopify via api/cli

Who is building products and services for agents? @agentmail and here.now are super interesting. Who else?

using tmux with moshi app on your phone, gives you access to the same sessions on your phone as if you were on your macbook. Same terminal, basically a port to your Mac on your phone. Be careful, it's more addictive than tiktok and instagram .

What's the most underrated tool that will 10x my productivity in Claude Code?

Running a DTC brand? Stop pulling fb/shopify reports every morning. Stop eyeballing metrics in ads manager. leannoku.com is a CLI and API that connects all your store data to Claude Code and your AI agents. All dots connected. Shopify, Meta, Google — you don't need to apply for each API separately. We handle that. Then just ask your data: "What's my blended ROAS across Meta and Google this week?" "Am I profitable today after ad spend, COGS and shipping?" "Which products have the highest repeat purchase rate?" "Did my Meta CPA spike overnight?" Save tokens. Save headache. One connection to all your data.

@zachmstuck I built a service app that gives you a simple API made for AI agents. Claude, ChatGPT etc. can just grab all your store data - orders, customers, products, whatever you need - DM if you're interested.

Shopify Sidekick is pretty useless. Was trying to get an exact number of customers for a social proof section on a landing page…this store has well over 300,000+ unique customers lol Each time I’ve asked a question, I always have to triple check math. Not sure why it’s so bad.

@johnrushx To have the best product - yes. Be public founder - why would ai care? They will judge your product, not you. Old approach i believe gonna matter to physical world, where ai can’t run 50 tests and compare you with your competitors

@thekitze open source?

fine fine i made my own email app i don't wanna pay monthly for another company to read my emails lol and openclaw is not the best at handling emails and giving me a peace of mind that my email is handled the first most obvious DUH feature i added is separation between human and automatic senders i can usually go to the automatic senders tab and just click archive all without thinking next up: llm classification and prioritization

Moshi + tailscale + tmux really unlocks the possibility of vibe engineering from your phone. I tried Termius. I tried different setups, but this seems to be the most simple and it just works. Basically, you have Claude Code that runs on your main computer or server, and those sessions are accessible on any device you have installed Tailscale. You download Moshi app on your phone, connect to your computer, and you have access to all your claude code sessions

The "agent-first" startup opportunity I can check Gmail as a human, refresh 500x, no limits. But access it as an AI agent? 250 API calls/day. Read 10 emails? That's 10 calls. Limits reached in 30min. Same with Twitter. I don't want your algo. I don't want to hunt for the bookmark icon in your UI. I want my agent to consume tweets for me Big Tech is building AI on one hand, but actively throttling agents with the other. Their products are still designed for eyeballs, not autonomous workflows. The next $10B company won't be "AI-enabled." It'll be agent-native built assuming humans are optional API limits = unlimited

OpenAI documentation doesn't support .md, com'on

@signulll Haha, yeah right

the fact that most of the world uses celsius is so fucking stupid. it’s objectively terrible ux for humans. zero meaning “water does a thing” is not how anyone experiences temperature. weather is a consumer interface. celsius is an implementation detail that’s actually in prod.

@seejayhess It is definitely a skill, but it's way easier to learn prompting skill than to learn someone's UI x.com/martynasmarch/…

I bet no one will use Cowork. Not because it isn't an incredible tool, but because people don't know how to use it. Claude Code is the ultimate prosumer tool. The 1% of 1% who can effectively leverage it in every which way can accomplish so much more than their pre-AI selves. But normal users don't really know what to do with it. I wouldn't even say this is a divide between professionals with an engineering background and those without one because I've seen many non-technical users become very proficient with Claude Code. The divide to me is a deeper skill somewhere between proficiency with AI and simple curiosity. There is a fundamentally different way to interact with a local agent versus a piece of traditional software. Feels like a push and pull thing. Most software you input data and then take actions to pull that data out in various ways. Dashboards, charts, spreadsheets, whatever. The software holds data and you pull from it. But agents you have to push. Push to find the right data, push to do the right thing, constantly nudge them along their way. This is a multi-step process that traditional software never had. Sure it can yield better results, but if you're bad at steering you never get there. Thats why I find my family to not be so impressed with AI. When I was home for Christmas I was talking about AI and in particular ChatGPT with my grandparents. My grandpa pulled out his phone, downloaded the app, and sent his first prompt. I forget what it was exactly but I instinctively stopped it and showed him how to improve it. The mistakes he made seemed so obvious to me but I think most people simply don't know how to prompt and use AI. This is the skill. I couldn't even really explain what was wrong besides "you have to give it everything to solve the problem" and he said "isn't it supposed to do that?" That one exchange kind of captures the whole problem. Technical people like us assume the barrier is the terminal. We think normal people are intimidated by command lines and code-looking interfaces. So we build prettier wrappers thinking thats the fix. Cowork is exactly this, Claude Code with a nicer UI and a friendlier name. But the real barrier was never the interface. It's the steering. It's knowing how to push. If you couldn't get value out of Claude Code, you probably won't get value out of Cowork either because the underlying interaction model is exactly the same. The tool was never the problem. The skill is.

Interfaces don't disappear, they just shift. Show user what's possible, give you a quick grasp. And AI takes it from there.

Form: 15 fields, dropdowns, validation errors. AI: 'Book me a flight to London next Friday' So is the future just a text box? No. Normal people don't know what to do with a blank input besides chat and search.

We've had 40 years to make forms work aaaand it still doesn't work. Here's what actually fixes it. Every dropdown, every text field, every "please enter your date of birth in MM/DD/YYYY format" - all of it exists for one reason: machines needed structured input from humans.