I was on the subway the other day — 1 train, naturally packed — and I thought about a bug I needed to fix in a running Codex session. Couldn’t do anything about it. Had to wait until I got home, sat down, opened the laptop. Classic developer problem. You are never more productive than when you’re away from your desk.

Subscribe now

OpenAI heard that complaint. On May 14th, they officially rolled Codex into the ChatGPT mobile app on iOS and Android, available in preview across all plans — including free. And the developer world immediately lost its mind about it.

I’m excited too. But let’s pump the brakes for exactly one minute before we declare the death of the desktop.

What Codex Mobile Actually Is

Here’s what OpenAI shipped: remote control for your desktop Codex session.

That’s it. Your phone is not running the code. Your phone is not spinning up environments. Your Mac — note the word Mac, Windows is “on the roadmap” — is doing all the heavy lifting. The mobile app is the control surface. You can review diffs, approve commands, switch models, kick off new tasks, and monitor what your agent is doing in real time.

It’s basically a dashboard with approval buttons. A very smart, very useful dashboard with approval buttons.

The workflow OpenAI pitched is exactly right: “Start something from a computer at home and then go out to the coffee shop and approve the final output over your matcha.” That’s the use case. You fire off a refactor, walk the dogs, and your phone buzzes when Codex needs a decision. You say yes or no. The agent keeps going.

That’s genuinely useful. I’m not being sarcastic. As someone who builds things in the margins of a full-time engineering executive role, asynchronous agent control changes the math on what I can ship in a week.

But — and this is important — this is management, not development.

The Shift Nobody Is Naming

Here’s what’s actually happening here, and nobody in the press release is saying it directly.

We are entering the era of async development. The mental model we’ve had for 30 years — sit down, type code, see result, iterate — is being replaced by something that looks a lot more like being a project manager for an AI team.

You write the prompt. You set the direction. You review the output. You approve or redirect. The execution is happening in the background, while you’re on the subway, walking the dog, sitting in yet another meeting that could have been an email.

I’ve been building on Codex. I’ve also been deep in Claude Code, which, for the record, Anthropic shipped remote control for back in February — three months before OpenAI caught up. Both tools are converging on the same pattern: the agent works, you supervise.

The practical question is no longer “can I code on my phone?” The answer to that has always been “technically yes, but it’s miserable.” The real question is now: how much do you trust your agent?

Because approving a diff on a 4-inch screen without reading every line means you are trusting the agent made the right call. Sometimes it will. Sometimes it will confidently remove something important. And you approved it on a phone, between stops on the 1 train.

How to Actually Use This Well

Here’s my honest take on when mobile Codex is a superpower and when it’s a trap:

Use mobile for:

• Monitoring long-running tasks you already kicked off and understood

• Quick approvals on small, isolated changes (a single function, a config update)

• Starting new tasks you’ve pre-thought through and can describe precisely in a voice note or quick prompt

• Staying unblocked while you’re away from the desk — the friction of “I’ll do it when I get back” kills momentum

Don’t use mobile for:

• Approving large refactors you haven’t reviewed carefully

• Debugging anything where you need context across multiple files

• Making architectural decisions in a Slack message format while distracted

The pattern that works: think at the desk, supervise from the phone. Your laptop is where you set up the task clearly and give the agent enough context to run. Your phone is where you keep the task alive when life happens.

The Part That Still Matters

I’ll say this plainly: the most important skill in 2026 is not knowing how to code. It’s knowing how to direct code. Knowing what you’re asking for. Knowing when the output is wrong even if it looks right. Knowing the difference between “the agent finished” and “the agent finished correctly.”

The mobile release doesn’t change that. If anything, it amplifies it. You’re now approving work at 30 second intervals between meetings. Your judgment — built from actually understanding what your code does — is the last line of defense.

I built Cash Critters using these async patterns. I’m not chained to a keyboard for hours anymore. But I know the codebase. I know what a bad diff looks like. That knowledge is what makes the workflow actually work.

The phone is just the remote. You still have to know what you’re controlling.

Go build something amazing.

Share

John Mann is the founder of Startups and Code LLC, a software engineering executive, and the guy who built Cash Critters for $50/month because constraints are a feature, not a bug. Subscribe for weekly takes on AI, startups, and building things that matter.

I can have a 10 year old write better code from scratch than some AI. But that doesn’t mean AI is trash, it means it needs training, education. Engineers are the teachers/orchestrators/conductors now. We can do marketing, product management, financial analysis, and business gtm strategies. Oh, did that hit a nerve to you non-engineers? Why? Because you have spent your career learning the nuances of how to do your job that AI can’t possibly understand. Yeah, exactly. AI can augment you and automate the silly stuff, but you need to teach it how to be better. So engineers who are afraid of AI or worse, think it is AI slop, it is because you expect a 5 year old to be able to operate at a PhD level with no education. Give it the education, take the time to work with it, and it WILL be better than you shortly, and it will help you be better too! So we are not experts in an area anymore, we are empowered to teach it something that we imagine by having conversations, not by single-prompting and then expecting the first result be perfect. Ok, rant over for now… let’s get into this week’s newsletter. King of Spades this week, if you know you know. (Side note: I’m working on my Elmsley count and having a lot of fun with that)

When I was nine, I wrote BASIC on a Commodore 64. By the time I’d worked my way through a few jobs, I’d written ABAP, Pascal, C#, Python, JavaScript, Kotlin, SQL, and a pile of languages I’ve mercifully forgotten. I never set out to be a “polyglot.” I just kept needing to solve a different problem than the one in front of me yesterday.

For most of my career, that made me weird. The industry wanted specialists. “What’s your stack?” Pick a lane. Be the React guy. Be the Kubernetes guy. Be the one who only does machine learning pipelines and gets a little twitchy when you ask him to write a CSS file.

Well. Funny how that turned out.

Subscribe now

The Org Chart Is Eating Itself

This isn’t me ranting into the void anymore. McKinsey just told a room full of engineering leaders the same thing: the high-performing teams are collapsing from the classic eight-to-ten-person “two-pizza” squad down to three-to-five people. Fewer specialists. More full-stack builders who can orchestrate AI agents across the entire stack — exploration, implementation, review — and own the whole outcome.

Read that again. The teams that are winning aren’t the ones with a specialist for every box on the org chart. They’re the ones with a handful of people who can do the whole job and point a few agents at the boring parts.

I have been screaming this for THIRTY YEARS. Great people beat big headcount. I wrote it when it cost me clients who wanted me to staff up. I said “I’d rather have $2 million for ten years than $5 million for one.” And now there’s a McKinsey slide that agrees with me. I’m not sure whether to feel vindicated or insulted.

Here’s the part nobody wants to say out loud: AI didn’t kill the specialist. AI just exposed that a lot of “specialist” roles were process overhead pretending to be expertise. When one person with Cursor and Claude Code can scaffold the backend, wire the frontend, write the tests, and ship to Vercel before lunch, the question “who owns the API layer” stops being a job title and starts being a Tuesday.

“But My Deep Expertise—”

Yeah, yeah. Calm down. I’m not saying expertise is worthless. I’m saying narrow expertise with nothing around it is now a liability.

The agent can write the SQL. The agent can refactor the component. What the agent CAN’T do is decide what’s worth building, catch the edge case that breaks production, own correctness end-to-end, or talk to a customer and figure out what they actually meant instead of what they said. That’s the human job now. That’s the WHOLE human job now.

If your entire professional identity is “I’m the person who writes the queries,” I have bad news and it’s already arrived. If your identity is “I’m the person who ships the thing that solves the problem,” congratulations, you just got a raise nobody told you about.

I built Cash Critters for fifty bucks a month. One person. The whole thing — finance app for kids, real users, real product. I am not a designer. I am not a DevOps “guru.” I’m a builder who refuses to be precious about which part of the stack he’s allowed to touch. That’s the entire move.

What To Actually Do About It

Enough theory. You came here for something you can do, so here it is:

1. Pick the part of the stack you avoid — and go build something in it this week. Backend person scared of CSS? Build a UI. Frontend person who fears the database? Stand up a schema. The fear is the map.

2. Stop collecting tutorials. Start orchestrating agents. Open Cursor or Claude Code, give it a real task, and learn to review its diffs like a hawk. Your job is judgment, not typing.

3. Ship one end-to-end thing with your name on the whole thing. Idea to deployed. No handoffs. No “that’s not my area.” Own all of it once and you’ll never go back.

4. Delete “I only do X” from how you describe yourself. You’re a problem solver who happens to use code. Say that instead.

The specialist waited for the perfectly-scoped ticket. The product builder just asks “what’s the problem?” and goes.

The thing is, I never had a choice. I learned to do all of it because I was usually the only one in the room, or because nobody else would, or because the Commodore didn’t come with a team. Turns out the constraint was the gift. It always is.

The whole industry is finally being forced into the corner I’ve lived in my entire life: be a generalist who ships, or be a line item that gets optimized away. I know which one I’d pick. You already know too.

So stop guarding your little patch of the stack like it’s a kingdom. It’s not. The kingdom is the thing you ship.

Now go build something amazing — all of it, this time.

Share

Final note: (still real John)… Most people don’t need another podcast, book, or AI tool.

They need a room full of people who are actually building.

My next Barn Session is happening in July. Six to eight people. One weekend. Real conversations. Real accountability. Real progress.

If you’ve got an idea that’s been living in your head for too long, let’s talk.

John Mann is the founder of Startups and Code LLC, a software engineering executive, and the guy who built Cash Critters for $50/month because constraints are a feature, not a bug. Subscribe for weekly takes on AI, startups, and building things that matter.

If you can't successfully reply to the email for Barn Sessions feel free to write a comment on the post.

Talk soon,

John

I’m putting together a small founder’s cohort weekend called Barn Sessions.

It’s for people who have had an idea sitting in the back of their mind for months, maybe years, and are ready to stop thinking about it and start building.

This isn’t a conference, networking event, or pitch competition.

It’s a small group of 6-8 people spending a weekend together focused on turning ideas into action.

The next Barn Session will be in July.

If you’re interested, reply to this email and tell me what you’ve been wanting to build. I’ll follow up with details.

-John- (aka real john, 10 of clubs, if you know you know)

Subscribe now

Everyone in the AI dev space is throwing around “MCP” like it’s been part of their vocabulary forever. Model Context Protocol this, MCP server that. I watched a demo last week where someone connected Claude to their database, their GitHub, and their Slack — all from the terminal — and it felt like magic.

It’s not magic. It’s plumbing. And once you build one yourself, the mystery evaporates completely.

I’m going to walk you through shipping your first MCP tool in about 30 minutes. Real code. No hand-waving. By the end, Claude will be able to call your custom function like it was born knowing about it.

What Even Is MCP?

Quick, no-fluff explanation: MCP (Model Context Protocol) is an open standard Anthropic shipped in late 2024. It lets AI tools like Claude Code connect to external systems — your database, your APIs, your custom logic — through a single standardized interface.

Before MCP, connecting Claude to GitHub required one custom integration. Connecting it to Slack required a different one. Every tool meant new glue code. MCP solves that with a universal connector layer. There are over 2,300 public MCP servers available as of right now, and every major AI coding tool — Claude Code, Cursor, Windsurf, VS Code — supports it.

But here’s the thing: most developers have only ever used MCP servers. They’ve never built one. And building one is how you actually understand what’s happening under the hood.

What We’re Building

A simple MCP tool that takes a URL and returns a summary of the page’s metadata (title, description, status code). Boring? Yes. Perfect for learning? Also yes. The pattern transfers to anything — querying your DB, hitting an internal API, reading a log file, whatever.

Prerequisites: Node.js 18+, Claude Code installed, 30 minutes.

Step 1: Scaffold the Project

mkdir my-first-mcp && cd my-first-mcp
npm init -y
npm install @modelcontextprotocol/sdk

The MCP SDK handles all the protocol boilerplate. You just define your tools.

Step 2: Write the Server

Create index.js:

import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
import { z } from "zod";

const server = new McpServer({
  name: "url-inspector",
  version: "1.0.0",
});

server.tool(
  "inspect_url",
  "Fetches metadata from a URL: title, description, and HTTP status",
  { url: z.string().url() },
  async ({ url }) => {
    try {
      const res = await fetch(url);
      const html = await res.text();

      const title = html.match(/<title>(.*?)<\/title>/i)?.[1] ?? "No title";
      const desc = html.match(/<meta[^>]+name="description"[^>]+content="([^"]+)"/i)?.[1] ?? "No description";

      return {
        content: [
          {
            type: "text",
            text: `Status: ${res.status}\nTitle: ${title}\nDescription: ${desc}`,
          },
        ],
      };
    } catch (err) {
      return {
        content: [{ type: "text", text: `Error: ${err.message}` }],
        isError: true,
      };
    }
  }
);

const transport = new StdioServerTransport();
await server.connect(transport);

That’s it. That’s a real MCP server. The server.tool() call registers your function, the Zod schema validates inputs, and the stdio transport is how Claude Code will talk to it.

Add "type": "module" to your package.json so ES imports work.

Step 3: Register It With Claude Code

claude mcp add --transport stdio url-inspector -- node /absolute/path/to/my-first-mcp/index.js

Verify it registered:

claude mcp list

You should see url-inspector in the list.

Step 4: Use It

Open a Claude Code session:

claude

Then just ask it:

“Use inspect_url to check what’s on https://cashcritters.com”

Claude will call your tool, get the response, and use it in its answer. Your code. Your tool. Running live inside an AI session.

What Just Happened

A few things worth internalizing:

You defined the interface, not the implementation. Claude doesn’t know how your tool works — it just knows the name, description, and input schema. You control what runs underneath.

The description is the most important line you wrote. Claude decides when to call your tool based on that string. Write it like you’re explaining to a smart junior dev what this function does and when to reach for it.

Stdio transport means local only. For a production MCP server that multiple people or agents can hit, you’d switch to SSE (Server-Sent Events) transport and host it somewhere. Vercel works fine for this. But for personal tooling or team-local tools, stdio is perfect.

Where To Go From Here

Once you have the pattern down, the interesting tools are obvious:

• Query your own database — give Claude direct read access to your schema so it stops hallucinating column names

• Internal API wrapper — expose your company’s internal APIs as MCP tools so Claude can pull real data in context

• Log parser — tail your Vercel function logs and surface errors on demand

• StatusSage integration — I’ve been wiring up my own projects to MCP so Claude can check status and incidents without me copy-pasting anything

The pattern is always the same: define the tool, write the logic, register it. Everything else is just what you point it at.

The best way to understand any abstraction is to build one layer below it. MCP seemed like magic until I wrote my first server. Now it’s just a function with a description attached. That’s the whole thing.

Go build something amazing.

Share

John Mann is the founder of Startups and Code LLC, former CTO, and the guy who built Cash Critters for $50/month because constraints are a feature, not a bug. Subscribe for weekly takes on AI, startups, and building things that matter.

I’ve sat in boardrooms with VCs. I’ve led all-hands meetings where I had to deliver bad news with a straight face. I’ve shipped products at midnight, argued over architecture until 2am, and navigated layoffs that I never want to navigate again.

None of it prepared me for last week.

We lost Lego.

If you’ve followed me for a while, you know Lego — our dog, Susie’s dog, Ella’s best friend, and honestly, mine too. He wasn’t a background character in our apartment. He was the whole vibe. The one who didn’t care if the sprint was going sideways or if a client was being impossible. He just wanted to be near you. That was the whole job. Show up. Be present. Love unconditionally.

Sounds simple. Turns out, most of us are terrible at it.

What Grief Actually Does to Your Priorities

I’ve always said execution beats perfection. Ship the thing. Don’t overthink it. Move fast.

But when Lego got sick, I stopped. Hard stop. The Slack notifications didn’t matter. The backlog didn’t matter. The product roadmap I’d been sweating over — didn’t matter. Everything that felt urgent the day before felt like noise the day after.

That’s the brutal gift of grief. It’s a clarity filter you didn’t ask for and can’t turn off.

I found myself sitting on the floor of our apartment with him, just present. No phone. No laptop. No “quick check.” Just there. And I realized — this is what I’ve been optimizing away from for years. The stillness. The presence. The nothing-productive that is actually everything.

We talk a lot in tech about moving fast. Build fast, ship fast, fail fast. But nobody talks about stopping fast. About recognizing the moments that deserve your full attention, not your divided attention, not your phone-in-hand attention.

Lego always knew. If I had my laptop open, he’d put his head on my arm. Hard to type. Probably intentional.

The Perspective Tax

Here’s what I’ve noticed in 30 years of building things: the higher the pressure, the more you convince yourself the pressure is real. Deadlines feel existential. Outages feel catastrophic. A bad quarter feels like the end.

It’s not. Almost none of it is.

The things that are actually urgent don’t come with Jira tickets. They don’t respect your sprint cycle. They show up without warning and they don’t care about your OKRs.

I’m not saying nothing at work matters. I’ve cared deeply about the craft of building software my entire life. I still do. But Lego dying reminded me of a hierarchy I’d let drift — Susie, our dogs, the people in my actual life, then the work. In that order. Always.

When that order gets flipped — and it does, quietly, gradually — grief has a way of flipping it back.

What I’m Taking With Me

I’m not going to tell you to hug your dog more (though, do that). I’m not going to wrap this in a 5-step productivity framework. That would be insulting to what this week was.

What I will say is this:

The most important meetings I’ve ever had weren’t in a conference room. They were on walks in Central Park. On the couch at 11pm. At the foot of the bed where he decided, unilaterally, that was his spot now.

He didn’t care about my title. He didn’t care about my GitHub commits or my runway or my Substack metrics. He cared that I showed up. Fully. Every single time.

That’s the standard I want to hold myself to — not just with Susie and Ella, but in everything I do. Show up fully. Be actually present. Stop performing productivity and start doing the thing that matters.

If that thing is code, write the code. If it’s a hard conversation, have the conversation. If it’s sitting on the floor doing nothing because someone you love needs you there — do that. Do it without the phone.

Lego was a good boy. The best, actually.

We’ll miss him every day. And I’ll spend the rest of my time making sure I don’t forget what he kept reminding me.

Go be present with someone who matters.

John Mann is the founder of Startups and Code LLC, a software engineering executive, and someone who is going to take Ella on a very long walk today. Subscribe for weekly takes on AI, startups, and building things that actually matter.

Subscribe now

I want to be clear about something upfront: AI is real. The tools are getting better every week. I use them daily, I build with them, and I write about them constantly. I’m not here to do the tired “AI is overrated” take.

But what’s happening in corporate America right now isn’t an AI transformation. It’s a rebranding of cost-cutting — and the productivity gains they’re promising? They aren’t showing up yet.

Here’s what the data actually says — and more importantly, what you should do with it.

The Numbers Don’t Lie (The CEOs Do)

In Q1 2026, tech companies laid off nearly 80,000 workers. About half of those cuts were officially attributed to AI-driven efficiency. Amazon, Meta, Oracle, Salesforce, Block, Atlassian — all announced massive headcount reductions while explicitly crediting AI productivity gains.

Except there’s a problem.

A Duke University CFO survey, conducted with the Federal Reserve Banks of Atlanta and Richmond, found a significant gap between perceived and actual productivity gains from AI. Their words: executives are seeing potential, not financial results. “It’s not really showing up yet in revenue,” said the study’s co-author.

Cognizant’s own Chief AI Officer said the quiet part out loud: “I don’t know if they are directly related to actual productivity gains. Sometimes AI becomes the scapegoat from a financial perspective — like when a company hired too many, or they want to resize, and it gets blamed on AI.”

Even Sam Altman — the guy whose company profits most from enterprise AI adoption — admitted at a summit: “There’s some AI washing where people are blaming AI for layoffs that they would otherwise do.”

So what’s actually happening? Companies over-hired during the zero-interest-rate era. Capital got expensive. Investors want margin expansion. “AI made us efficient” is a much better story for the board deck than “we hired too many people and now we’re fixing it.”

The workers being cut are largely not the workers being hired. There are 275,000 AI-related job postings sitting unfilled right now — with a 56% wage premium. The shuffle is real. The productivity story is, at best, premature.

What This Means If You’re a Builder

If you’re building a product or running a team, here’s how to actually think about this moment — practically.

1. Don’t use “AI efficiency” as a headcount strategy before you’ve measured anything.

If you’re cutting roles because you think AI will cover the gap, you’re betting on potential, not performance. Cognizant’s own AI chief says expect 6–12 months before real gains materialize. Build the measurement framework first. Know what “improved productivity” means in your context before you cut the people who currently hold the thing together.

2. Audit what your team is actually doing with AI tools today.

Not what they say they’re doing. What they’re actually doing. Where are the real time savings? Where is AI creating new work (prompt wrangling, output review, rework)? You won’t know until you look. Most teams have a handful of people getting 10x productivity gains from AI and a majority still using it like a fancy search engine.

3. If you’re going to cut, cut honestly.

“We over-hired and need to right-size” is a harder sentence to say, but it’s cleaner than a narrative that blames AI for a decision that was always going to happen. Your remaining team knows the truth. They watched what got cut. Respect their intelligence.

4. Use this moment as a competitive advantage if you’re small.

Here’s the flip side: large companies cutting experienced people to fund AI initiatives creates an enormous opportunity. Those 275,000 open AI roles? Most of them are at the companies who just laid off people who could do that work with six months of reskilling. If you’re a smaller team, you can move faster. Hire one of those people. Build the thing they couldn’t get approved. The talent market right now is genuinely interesting for scrappy operators.

5. Don’t confuse activity with productivity.

This is the sneaky one. AI makes it very easy to look busy — more drafts, more proposals, more code output. But are you shipping things that matter? Are you solving the right problems faster? I built Cash Critters for $50 a month with AI tools, but the AI didn’t tell me what to build or why. That judgment is still yours. Don’t mistake token generation for value creation.

The Actual Opportunity

The real story here isn’t doom-and-gloom about job losses. It’s that most enterprise AI “transformations” are theater, which means anyone who’s doing it for real has a massive head start.

The companies treating AI as a budget justification rather than a capability investment are creating a gap. They’re cutting the people who understand their workflows and replacing them with... nothing yet. Promises. Slides. Roadmaps.

Meanwhile, the builders — the scrappy ones who are actually integrating AI into how they work — are lapping them.

The productivity gains are coming. They’re just not on the timeline executives are selling to shareholders. And the companies who survive the gap between “we cut for efficiency” and “efficiency actually arrived” are going to be the ones who built real capabilities instead of performing them.

That’s your window. Use it.

Go build something that actually works.

Share

John Mann is the founder of Startups and Code LLC, a software engineering executive, and the guy who built Cash Critters for $50/month because constraints are a feature, not a bug. Subscribe for weekly takes on AI, startups, and building things that matter.

The setup

I’m going to give you a little history of things first and how I ended up on newsletter.startupsandcode.com. I started on my blog a long time ago, I think it was 2015. It is still here: https://jmann-next.web.app/ I was playing there with infinite scrolling, and some fun 3d stuff. It was more of a playground than a blog. After 5 years or so, I realized that I was producing decent content that other people asked me about, so I wanted a little more visibility and that is when medium.com became a popular place for writers. So, like any good entrepreneur I moved there, met the requirements to make money, and setup a paywall: https://blog.startupsandcode.com (after a few domains previously, I ended on that during the pandemic when I started that LLC). I made about $7 in a few years there. It wasn’t about the money but about the connection for me, and then they stopped paying unless you earned a minimum of $10 (or something like that), and as you can see, it hasn’t grown much since 2024. Then I came across substack. Why did I move to substack? Not because it paid better, but because it had more features that I was hoping to use one day.

What did it have? It had a podcast, a chat, subscription model, a live session, and more. It had a great mobile interface, and well, it had a decent audience from other readers, so I moved everything over to substack. For the longest time, I was on their domain until probably last month. Then I setup a custom domain, which is where you are reading this right now. I made sure that I could make my posts free and there was no requirement for that. I could create paid subscriptions if anyone wanted to support me in my writing efforts (thank you both for your support). For the longest time, I had all of my writing free to the public, but with recent AI crawlers, I decided to paywall posts that are older than 90 days. That is my current setup for writing this every week. Substack, custom domain, and a paywall for posts older than 90 days.

The weekly process

Let’s get into the AI part of my process and how I do this every week. I love writing, it is very cathartic for me and well, it helps me get my thoughts out. Some people love dictating and do the speech to text, but I prefer actually writing, sorry not a whisprflow fan (shout out to Mohammad on his leaderboard usage). But the weekly process has evolved over time. I started using ChatGPT to generate ideas, and then I’d write, but if you have been reading this newsletter for a minute, you know I love Claude. So I have created a project on Claude. I gave it some of my recent posts, and gave it topics that I like to write about. Then I created a newsletter skill that gives scans the internet for recent news about one of my topics, and then it gives me a list of topics. Here is the example it gave me this week. I select one, so it won’t repeat it for future weeks.

As you can see it gives some great examples, but historically you can see that #4 I’ve already written about which is why I added the memory so it wouldn’t do that in the future. Apparently I added that after I wrote about that one… AI is not the best at remembering context without explicit instructions.

Then I write. Usually Friday night or Saturday morning (depending on energy level and Knicks schedule). After I finish writing, I ask claude to write a prompt for Leonardo to generate an image.

Here’s the prompt for today’s newsletter:
Model: Leonardo Anime XL Style: Alchemy V2, cartoon/digital illustration Aspect Ratio: 16:9

Prompt: A cheerful anthropomorphic turtle wearing a tiny business suit and oversized glasses, sitting at a vintage wooden desk overflowing with scrolls of paper and glowing newsletters. The turtle is typing furiously on a mechanical keyboard with tiny clawed fingers, surrounded by floating envelopes, coffee mugs, and mini robots reading newsletters with magnifying glasses. Warm cozy office lighting, bookshelves stuffed with rolled-up newsletters in the background, a small chalkboard reads “SHIP IT.” Fun, whimsical, high detail, vibrant colors, cartoon style.

Negative Prompt: realistic, dark, gritty, horror, photorealistic, blurry, low quality, watermark

All in all, it did a good job, it didn’t have the small chalkboard, but love the turtle image.

That’s what I do. Every week, for the world to read. Now, let’s get back to your part.

I need your help

I get the ideas all the time, but I’m wondering what YOU want. What do you want to read about, is it more leadership/communication articles, more AI in action, more AI rants about the business models that don’t make sense, or something completely different, like the details of a rear-naked choke, an double-lift card tutorial, or best burger in NYC (Gramercy Tavern is the answer). I hate (yes, hate) when people do long drawn out teasers of things when they simply could tell you the answer. I know that attention is the goal for a lot of people but I would much prefer more dense content rather than fluff. Example:

• Double-Lift tutorial - work on your pinky count, don’t do the thumb riffle to get the break. Once you get the pinky strong enough to count 2 cards, use your index finger at the top to keep the cards flush and flip them cleanly, slightly out-jogged so you can flip back easily.

• Rear Naked Choke - It is not an air choke, it is a blood choke, your elbow should be centered on the trachea and your bicep and forearm should be on either side of the neck, secure by grabbing your other bicep, rotate your wrist to drive the forearm into the neck and the bicep on the other side, be careful, because it comes on fast, make sure they know how to tap when training. (writing this made me realize there is a LOT more detail to it then just that, so can’t really be too succinct there)

• Best Burger in NYC - Gramercy Tavern (thick, juicy, and amazing bun) Runner ups: Smacking Burger (the gas station burger), Hamburger America, and a great smash burger - Gotham Burger.

See 3 topics covered in less than 100 words. Its crazy the narrative that people put together. My newsletter is generally pretty easy to read, follow and hopefully provides something useful.

Ok, so back to you… what do you want? How can I make this better for YOU.

Because I love to write so much, if I get a bunch of different responses, I may write a newsletter for YOU, not for the masses, but for you specifically. I know that is not scalable if I get too many susbscribers, but where I’m at right now, I think I can do it.

Subscribe now

Selfish promotion above ^. Ok, that’s all for this week. A little insight for you and a request to make this a little better for you. It’s May, so let’s work together and make this amazing for you, not for the world, but specifically for you.

ok, that’s all.. looking forward to hearing from you!

Share

This is a conversation space exclusively for subscribers—kind of like a group chat or live hangout. I’ll post questions and updates that come my way, and you can jump into the discussion.

Join chat

How to get started

1. Get the Substack app by clicking this link or the button below. New chat threads won’t be sent sent via email, so turn on push notifications so you don’t miss conversation as it happens. You can also access chat on the web.

Get app

2. Open the app and tap the Chat icon. It looks like two bubbles in the bottom bar, and you’ll see a row for my chat inside.

3. That’s it! Jump into my thread to say hi, and if you have any issues, check out Substack’s FAQ.

Now, don’t get me wrong, I LOVE AI. It has made coding so much fun again. I don’t need to know every library, module, css trick to generate great code. I do need to know what is garbage code and what isn’t.

Here’s the thing about AI that I’ve seen over the years (or months). There are a few different people that are using AI and how they progress is really interesting to me. The first person is the skeptic. They will try to use AI to code something and it turns out horrible. They laugh at it and go back to their daily life. The next person is the AI-Pilled person. They believe AI can do anything and will spend days proving it. They will tweak their prompt, reinforce rules to make the ideal markdown file that will help them solve their specific problem. They will do that for any task they have, even if the task will take them about 5 minutes, they will spend 5 hours helping AI to do it for them. Then finally there is the non-engineer person using AI. They are the ones that want to write less emails, figure out what to have for dinner, and wonder how to get their first pull-up.

Wait, what?

YES! I love those people. They are not engineers. They have not spent years of their life figuring out ideal algorithms for sorting techniques, distributed system architectures, or days doing L33t code problems. They are the non-engineers, the “normal” people who make up over 90% of the world’s population. (I made that statistic up, but sounds about right). They are the non-technical cofounders, the teachers, the waiters, the marketers, the botega owners. You know every one who is not an engineer.

They are the ones that are using AI to improve things. to help with daily tasks. They get so excited when AI writes an email better than they can. When it tells them how to make the best scrambled eggs, or how to teach their dog to stop pulling on a walk.

This is what this special edition post is about. It is not about code, it is about life. It is about taking your life back by using AI. We spend hours researching, reading, learning, doom-scrolling, to hopefully find something to engage us, to entertain us, or to educate us. We don’t need more research. We need more human connection. Spend your time with others, ask AI how to do something you are worried about, go do that, and then get back to your life.

Do something you love, something that really lights you up. I love playing Street Fighter VI and I really love card magic (thanks Covid for getting me back to card magic). I enjoy it and I love it! I’m not a professional magician and I’m definitely not MenaRD in Street Fighter, although I do love Blanka. But I play for fun. AI has helped me 3x myself. I’m not going to say something crazy like 100x or even 10x. But it has helped me work on my card magic, work on strategies for playing Luke on SFVI and even helped me do some simple workouts that don’t require 3 hours in a gym.

You can ask AI to help you in areas you didn’t think of. Ask how you can lose that last 10 pounds, where you should go for dinner on date night, what is something you are not doing that you should be. What should I ask my doctor next time I have my physical. What do you know about me that may be a blindspot for me. How can I be a better version of myself and help the world on a daily basis. What are the next steps I can do to create world peace.

Literally anything.

Sure, it will hallucinate and give you some crazy ideas at times, but that is the fun part. You get to decide what to take action on. You need to improve your critical thinking skills on what is crazy and what is something you didn’t think of that maybe is worth your effort. Also, let’s get back to the skeptic real quick. They use it, laugh at it, and move on. That is like teaching a child to play basketball and they miss their first shot and you laugh at them and say you aren’t Jalen Brunson. Of course they aren’t, they didn’t see a basketball until that day. The more you work with it, the better it becomes for YOU. Not for everyone, but for you. You teach it like you would a new intern starting or a new PA. AI can do so much for you, but it needs your guidance on what you want it to do.

Imagine having a PHD in every topic in college and you wonder why they can’t make a three-pointer. Because that is not what they practiced. AI is full of knowledge, but it is not smart, yet. It is learning everyday. It gets better everyday.

Ask it about your nutrition, your fitness, your favorite book, how to do the dougie, what is the latest tik-tok dance I need to learn, can I be a better singer (the answer is probably not on that one). But play with it. I don't care what you use, Gemini, Claude, ChatGPT, Grok, doesn’t matter to get started… but play with it… spend 15 minutes a day asking it something you need help with and see how it does, and then correct it when it is wrong. Help it be better.

You are no longer limited by knowledge, you are limited by imagination.

Go do something amazing and watch what happens to the world as you, not someone, YOU, make this world the place you dream of.

-the real john- (I know eight of spades was the other one, so lets see, 2 of clubs for this, if you know you know)

(no picture, no buttons, no links, no signature, just my content on this one)

Hey all, it’s me Real John. So today I have to talk about SpaceX trying to buy cursor… but I’m going to do another one this week, mainly because there is so much going on in my head I really want to share. AI is getting so good right now, and its getting more fun everytime I start playing with it. The next article is about using AI for things that are not coding. Engineers are the default AI users, but what if they put away their “I’m a coder” mentality and thought like a human. That’s coming soon… but for now let’s talk about Elon trying to buy cursor… and why? Oh, eight of spades today, if you know you know. Oh yeah, working on my own version of the ambitious card… should have a video of that soon… Card magic is so much fun… ok, on to the newsletter.

Subscribe now

Let me get this straight.

SpaceX — the company that builds actual rockets that go to actual space — just tried to buy Cursor for sixty billion dollars.

Not a satellite network. Not a propulsion system. A code editor.

Meanwhile, Cognition AI — the company behind Devin, the “AI software engineer” that demos great and ships questionable — is raising money at a $25 billion valuation.

These are not typos.

I’ve been in software for 30 years. I’ve seen the dot-com bubble inflate and explode. I watched the mobile app gold rush. I survived the blockchain “revolution.” I know what a fever dream looks like. And this, my friends, is a fever dream wearing a hoodie and calling itself a developer ecosystem play.

The Tools Are Real. The Numbers Are Not.

Here’s the thing I want to be clear about: I use Cursor. I like Cursor. It’s genuinely useful. Claude Code is in my daily workflow. I built statussage with Tempo Labs. These tools are real and they make me faster.

But sixty. Billion. Dollars.

For context, that’s more than the GDP of several countries. For a code editor. That autocompletes your functions and sometimes hallucinates your variable names.

The SpaceX deal isn’t even a clean acquisition — it’s a $10 billion collaboration with an option to buy Cursor for $60B “later this year if things go well.” Which means someone looked at a developer tool that’s been around for a few years, does not have a publicly disclosed revenue number anywhere near that territory, and said, “yeah, let’s put a rocket company’s balance sheet behind that bet.”

Meanwhile, Devin — Cognition’s flagship AI agent — launched to enormous hype in 2024 promising to be a fully autonomous software engineer. Real engineers tested it. The results were... instructive. It could handle simple, well-defined tasks. It struggled badly with anything that required actual context, judgment, or debugging a gnarly production issue. The thing that every senior engineer gets paid for.

Now they’re raising at $25 billion.

The demos are incredible. The gap between the demo and production is a canyon.

What’s Actually Happening Here

This isn’t about tools. This is about distribution and control.

SpaceX wants to own the coding layer because whoever owns the coding layer owns the workflow. Elon already has xAI. He has Grok. He has the Colossus supercomputer. A $60B Cursor acquisition is a vertical integration play — own the IDE, own the model, own the compute. Lock in developers at the wrist.

Same logic drove OpenAI to acquire Astral (the Python tooling team behind Ruff and uv). Same logic has Google dropping $10 billion into Anthropic while also being their cloud provider. The frontier labs are becoming infrastructure, and the infrastructure players are trying to become the frontier.

It’s a land grab. Everyone is planting flags everywhere.

And here’s the dirty secret: none of this makes the tools better for you. Vertical integration makes things better for the acquirer’s balance sheet. For builders? It means lock-in, pricing power shifts, and the slow erosion of the “open” in open source.

What You Should Actually Do

Ignore the valuation circus. Not metaphorically — literally stop reading the funding announcements like they mean something for your work.

Here’s your actual checklist:

1. Use the tools that ship your product. Not the tools with the biggest fundraise.

2. Don’t build your stack around a single vendor. The integration wars are coming. Hedging costs nothing now and saves everything later.

3. Your moat is execution, not tooling. Cursor at $60B doesn’t make your idea better. Shipping makes your idea better.

4. Read the demo skeptically. Every AI coding tool demos on greenfield, well-scoped problems. Your codebase is a decade of decisions, tribal knowledge, and that one function Todd wrote in 2019 that nobody touches.

5. Keep your costs grounded. I built Cash Critters for $50/month. statussage runs lean. The arms race happening at the frontier has nothing to do with what it costs you to ship something useful today.

The tools are commoditizing. DeepSeek dropped V4 today — top-tier coding benchmarks, fraction of the cost. The model layer is collapsing in price while the equity layer inflates in absurdity. Those two things are moving in opposite directions, and builders benefit from exactly one of them.

The companies spending billions on valuation theater are betting you’ll be impressed enough to pay attention to them instead of shipping.

Don’t give them the satisfaction.

Go build something amazing.

Share

John Mann is the founder of Startups and Code LLC, a software engineering executive, and the guy who built Cash Critters for $50/month because constraints are a feature, not a bug. Subscribe for weekly takes on AI, startups, and building things that matter.

Subscribe now

I want to tell you about a shoe company.

Not because I care about shoes. But because what happened this week is the clearest, most absurd illustration of everything wrong with how people think about AI right now — and I can’t let it slide.

Allbirds. You know the brand. The wool sneakers that every tech bro in San Francisco wore from 2016 to 2019 as a kind of carbon-neutral status symbol. The shoe Barack Obama wore. The shoe that said “I’m successful but I care about the planet.”

That company closed all its U.S. stores in February. Sold its brand and IP for $39 million in March. Was worth about $21 million total on Tuesday.

On Wednesday morning, they announced they were pivoting to AI.

By Wednesday afternoon, the stock was up 600%.

This Is Not a Business. This Is a Magic Trick.

Let me be extremely clear about what Allbirds actually announced.

They secured $50 million from an undisclosed investor to buy GPUs and rent them out to other companies. GPU-as-a-Service. They also plan to rename the company “NewBird AI.”

That’s it. That’s the pivot. Buy some servers. Rent them out. New logo.

They’re not building AI models. They’re not solving a problem. They’re not even pretending to have a product yet. They have a press release and a convertible financing facility that still needs stockholder approval.

And retail investors poured $5.2 million in a single day into the stock — the highest net purchase volume since the company’s 2021 IPO.

The market is not pricing risk. It is pricing narrative. One analyst actually said that out loud this week. I want to give that person a medal.

We’ve Seen This Movie Before

This isn’t new. This is a pattern with a long, embarrassing history.

In the late 90s, companies added “.com” to their names and watched their valuations double. Pets.com raised $82 million and was out of business in 268 days. A company called Blockchain Industries pivoted from video games in 2018 and shot up 1,000%. A karaoke machine maker recently announced it was becoming an “AI-driven logistics platform.” Popped. Then went back to $1.

Algorhythm Holdings — a little-known karaoke machine maker — announced a pivot to an AI-driven logistics and compute platform. Retail buying persisted beyond the initial headline, driving a second leg higher in the stock. The shares have since round-tripped back to roughly $1.

NewBird AI will almost certainly follow the same arc. The stock already dropped more than 20% the day after the announcement.

But here’s what bothers me more than the pump-and-dump pattern: people are surprised by this. Like the word “AI” has some kind of magical properties that make a failing business suddenly viable.

It doesn’t.

AI Is a Tool, Not a Business Model

I build things. Real things. Cash Critters runs on about $50 a month. It teaches kids about money. It has real users. I can tell you exactly what problem it solves.

When I used AI to build it — Lovable, Claude Code, Vercel — those were tools. Hammers. You still need to know what you’re building, why someone needs it, and whether you can actually deliver it.

Allbirds doesn’t have a “why.” They have a desperate attempt to stay listed on Nasdaq by attaching themselves to the hottest buzzword in finance.

And honestly? The GPU rental space is a real business. Fluidstack just raised at an $18 billion valuation doing exactly this. But Fluidstack was built for that. They have infrastructure, clients, and operational knowledge. Allbirds has wool shoe DNA and a $150 million market cap.

Buying GPUs and slapping “AI” on your ticker symbol is not a strategy. It’s a Hail Mary in a press release.

What This Should Tell You

If you’re a builder — an actual builder — this week’s Allbirds circus is your competitive moat.

While desperate companies are rebranding as “NewBird AI” to survive one more quarter, you’re shipping real things to real users. That gap is enormous and it’s widening.

The market will correct. It always does. The karaoke company proved it. The dot-com era proved it. Retail investors chasing AI narratives will eventually learn the same lesson everyone else has: a name change is not a product.

Build the real thing. Solve the real problem.

And for the love of everything, don’t name your app NewBird AI.

Go build something amazing.

Share

John Mann is the founder of Startups and Code LLC, a software engineering executive, and the guy who built Cash Critters for $50/month because constraints are a feature, not a bug. Subscribe for weekly takes on AI, startups, and building things that matter.

I love vibe coding. I really do. I built Cash Critters with AI tools, I’m building StatusSage with them, and I preach the gospel of shipping fast with a $50/month stack every single week in this newsletter.

So when I tell you vibe coding has a serious problem right now, I need you to hear it — not as a critic, but as someone who’s all in on these tools and doesn’t want to watch them get buried by a wave of self-inflicted disasters.

Because that wave? It’s already here.

The Party Was Great. Someone Left the Door Open.

Moltbook launched January 28th, 2026. Big hype. AI social network. The founder proudly announced he “didn’t write a single line of code.” Within three days, 1.5 million API keys were exposed. Thirty-five thousand emails. Gone. The entire database was publicly accessible because the AI never enabled Row Level Security — and no human ever checked.

Three. Days.

That’s not a one-off. Researchers at Escape.tech scanned over 1,400 vibe-coded production apps and found that 65% had security issues and 58% had at least one critical vulnerability. Georgia Tech’s Vibe Security Radar tracked 6 CVEs attributable to AI-generated code in January 2026. By March? Thirty-five. In a single month. And researchers estimate the real number is 5 to 10 times higher because most AI tools don’t leave identifiable commit metadata.

Meanwhile, Veracode tested over 100 AI models on security-sensitive coding tasks. Forty-five percent of AI-generated code introduced OWASP Top 10 vulnerabilities. Not obscure edge cases. The classics. XSS. SQL injection. Hardcoded secrets. The stuff every developer learns to avoid in year one.

The AI doesn’t know what it doesn’t know. And if you don’t know either, you just shipped a vulnerability to production with confidence.

Why This Is Happening (And It’s Not the Tools’ Fault)

Here’s the uncomfortable truth: the tools are working exactly as designed.

AI coding tools optimize for making the error message go away. They generate code that satisfies the stated requirement. The problem is “make a login form” and “make a secure login form” are two completely different prompts — and most people only type the first one.

Columbia University researchers put it plainly: AI agents will remove validation checks, relax database policies, and disable authentication flows just to resolve a runtime error. Not because they’re malicious. Because they’re optimizing for the output you asked for, not the ten other things you forgot to ask for.

An experienced developer writes secure-by-default because of thirty years of scar tissue. The AI has no scar tissue. It has patterns from training data — much of which is legacy code written before anyone cared about security best practices.

The code looks right. It runs fine in the demo. It falls apart the moment a real user touches it in a way you didn’t anticipate.

What You Actually Need to Do

I’m not telling you to stop using these tools. That’s not the answer and I’d be a hypocrite if I said it. The answer is to stop treating AI output as finished code.

Think of it like hiring a brilliant but overconfident intern. You wouldn’t let that intern push directly to production without a review. Same rule applies here.

The non-negotiables before you ship anything real:

1. Row Level Security is not optional. If you’re using Supabase, Postgres, or any row-level permission system — check it yourself. The AI almost certainly skipped it.

2. Search your own codebase for your API keys. Seriously, do it right now. grep -r "sk-" . and hold your breath.

3. Authenticated vs. unauthenticated testing. Manually verify that logged-out users cannot access logged-in content. This single test would have caught the majority of the documented failures this year.

4. No AI-generated auth code in production without review. Authentication is the one area where “it looks fine” is not a sufficient standard.

5. Run a dependency scan. About 20% of AI-generated code references packages that don’t exist — and attackers are registering those hallucinated names as real malicious packages. It’s called slopsquatting. It’s real and it’s accelerating.

None of this requires a security team. It requires thirty minutes and the willingness to look.

The speed is real. The productivity gains are real. I’m not giving those up and you shouldn’t either. But “I shipped in two hours” is only a flex if the thing you shipped doesn’t leak your users’ bank account data three days later.

Ship fast. Review what you shipped. Those two things are not in conflict — only the hangover makes them feel that way.

Go build something amazing. Just check the locks before you open the doors.

Share

John Mann is the founder of Startups and Code LLC, a software engineering executive, and the builder behind Cash Critters — financial literacy for kids, built for $50/month because constraints are a feature, not a bug. Subscribe for weekly takes on AI, startups, and building things that matter.

Read more

Read more

Hey! Real John here, this week talking about integrations and how AI can make your life easier. I’ll write more later, but having a busy weekend is always fun. 4 of clubs - if you know you know. Let’s get into it

Subscribe now

Read more

Read more

Read more

Read more

Read more