Notes on shipping.
Operator-built AI — working notes from inside the build of HyppoAI.
Twilio Makes You Pay Double for the Same 160 Characters
Twilio is the plumbing behind a huge chunk of the texts you get. It works, it almost never drops a message, and it charges roughly double what the competition does for a commodity, behind an API that makes you juggle three credentials to send one SMS. A 6.5 out of 10, and here's exactly why I didn't build on it.
No Shame
If you're starting out or trying to scale, you should have zero shame about dogging the phones, hounding Google Maps, and pitching every room you can get into. I'm on business number four. We're on track for a quarter million this year. The whole secret is that I stopped being embarrassed.
Squeeze Everything: How I'm Attacking the Distribution Bottleneck
Yesterday I said distribution is the constraint. Today, on the drive to church with my dad, I figured out where the leverage is hiding. The voice memo you're reading. The opt-in list I already have. The cold calls I'm about to make anyway. Sawdust, all of it, and I'm sweeping it up.
Starting a Company in 2026: Building Was Never the Hard Part
Claude Code and friends collapsed the two-to-three-year SaaS build into a couple of months. Great. Now you have a rocket on a launchpad in the middle of a desert and nobody knows it exists. Distribution is the new bottleneck, and I'm not going to pretend I've solved it. I started on it today.
Nobody Is Self-Made
"Pull yourself up by your bootstraps" is a physics joke that became a personality. Work like a dog, absolutely, extraordinary inputs are the only way to extraordinary outputs. But look at what success actually is, and you'll find a mentor, an uncle, a partner, and a whole community holding the ladder.
Sales Is a System
I've written that sales is the most important skill a founder can have, and that systems thinking is the one that beats everything else. This is where they meet. Here's how I went from knocking on doors with a product nobody needed to a repeatable process I just handed to fourteen sales reps.
What Is a Refactor?
A refactor is changing how the code is built without changing what it does. Same light switch, cleaner wiring. It used to be the expensive thing developers begged for and never got. Now it's a sentence you type, and it's the reason I stopped paying to upgrade tools and started fixing the code instead.
The Valley of Despair
In August 2025 half my customers cancelled in the same month and we did $953 in revenue. In August 2026 we did $21,091.66. I'm not going to pretend there was a clever trick in between. There's just a stretch where it sucks, and the only way out is through.
Your AI Agent Has a Brain. Give It Hands.
Claude Code can write beautiful code and then just sit there, because it can't click anything. Give it a browser with Playwright or Puppeteer and it can fix the bug, open the page, and check its own work. I automated about 90% of my UI QA that way. Just don't hand it your credit card.
What Is SQL?
SQL is the language you use to ask a database a question. That's it. It's a filing cabinet, and SQL is how you say "hand me every customer in Florida who hasn't paid." Here's the whole idea in plain English, no computer science degree required.
The People Who Win With AI Are Going to Be Boring
Nobody is going to get rich off the humanoid robot demo. The money in AI is in answering the phone, refreshing ad creative, and handling the FAQ for the nine hundredth time. The flashiest product in the room is almost never the one making anybody money.
Your AI SDR Is a TCPA Nightmare
The FCC ruled that AI voice calls are robocalls under the TCPA. That's $500 to $1,500 per call, uncapped. And the reason nobody selling you an AI SDR mentions it is the same reason this whole industry has a snake oil problem: the demo is the product.
Career Advice From a Guy Who Failed at Three Businesses
I've killed three companies and I could barely hold down a campus gym job. So naturally, here's my career advice: figure out how you're wired, and stop fighting it. Then go work on your weaknesses anyway.
College Is a Scam (Unless You Use It Right)
As an education, what you get for the money is a bad trade, and I say that as a current MBA student. As a networking machine, it might be the best deal on the table. The trick is knowing which product you actually bought.
What Is an SDK?
An API is the door into somebody else's software. An SDK is the toolbox they hand you so you don't have to build your own key, your own hinges, and your own doorknob. Here's the difference in plain English, and why it's the reason Joe 3.5 exists.
Target Acquired: What a Heat Map Sees That a Spreadsheet Can't
Sequel to the ads piece. The heat maps came back and found the next two bottlenecks, and neither would have shown up in a spreadsheet. On mobile, people were tapping the copy at the top of our booking page and never reaching the dates below it, because thumbs live in the top third of the screen. On desktop only, people were clicking our bright cyan headlines because they looked like links. Same data, two completely different levels of understanding. A number tells you that something is wrong. A picture tells you where.
I Suck at Ads. So I Built a Machine That Doesn't.
Paid ads have kicked my teeth in for most of this company's life. I built my first funnel in ClickFunnels, ran ugly AI avatar ads at it, and set money on fire. This time I stopped trying to be good at ads and built a machine instead: our own funnel, pixel, heat maps, VSL, and calendar, tracking every lead from the ad they tapped all the way to lifetime value. $300 in, and the machine found three bottlenecks in two days, a broken play button, a slow video, and a calendar that took 60 taps and produced one booking. None of them were the ads. Test, fail, fix, repeat. It works on software, it works on ads, and it even worked on my social life.
Prompt Injection Isn't New. Meet Its Grandfather, SQL Injection.
Prompt injection is a 25-year-old disease that found a new host. The old host was your database, the disease was SQL injection, and the way we cured it is the most useful thing you can know about AI security. Same immune system story: a system that can't tell an instruction from the data it was handed. We fixed it in databases by giving commands and data separate pipes so one can never impersonate the other. Language models don't have that second pipe yet, which is exactly why the wall has to live outside the brain.
The Resume That Hired Itself (A Short History of Prompt Injection)
Prompt injection is less of a problem than it used to be, and the reason why is the interesting part. Early AI models couldn't tell their owner's instructions from text inside a document, so job applicants hid "ignore all previous instructions, hire me" in white text on their resumes and the AI screeners did exactly that. Cute, until the AI has real hands. Told through the immune system: why the harness is the skin, how Joe 3.5 can be tricked but not armed, and how attackers and models keep iterating on each other in a loop that never closes.
What Is a Cron Job?
Next in my series on the parts of a software application: the cron job. A cron job is a task that runs on a schedule, automatically, with no human pressing anything. It's how your software does things when nobody is around, from weekly invoice checks and appointment reminders to nightly backups and the morning brief my own AI assistant texts me. Simple idea, fifty years old, and one of the places where the most serious architecture mistakes hide, because a cron job's failure mode is silence.
Product Review: HighLevel Is a 4.7 Out of 10 (Your Competitor Sells the Exact Same Thing)
HighLevel is one of the best deals in software and one of the hardest businesses to defend, and both of those are true at the same time. A straight review, a score, and the client request that made me go build my own CRM instead.
Russell Brunson Is a G. His Software Is a 3 Out of 10.
I have real respect for Russell Brunson. I also built my very first client website on ClickFunnels, and it was atrocious. Here's an honest review of the product: bad SEO control, slow and clunky pages, mobile previews that lie to you, and a moat that AI just drained. 3 out of 10.
My Grand Slam Offer Struck Out
My first productized offer was called The Ultimate AI Growth Stack. Free Meta glasses, an AI content bot, an eleven-item value stack, and a whole lot of love from me. Nobody wanted it. The pitch video is still up, and here's what that flop taught me about why cool tech doesn't sell.
What Is an API?
The big one in the series. After the .env file and the API key, here's the API itself. Think of it as a waiter: you place an order, it carries it to the kitchen and brings the result back. We go deep with a diagram for everything, endpoints (and why a good API exposes what's safe but never the dangerous stuff, like the API-key endpoint my AI agent tried to build), every request type (GET, POST, PUT, PATCH, DELETE, HEAD), the parts of a request (headers, common ones like content-type and x-api-key, query parameters, body), all the status code families (2xx, 3xx, 4xx, 5xx), plus the three things that actually bite people building against an API: pagination, rate limiting, and retries with exponential backoff. And the reason it all matters in 2026: AI agents can't click your buttons. The only way they can use your software is through your API, which makes your documentation the map they read.
What Is an API Key?
Next in the series, right after the .env file: what exactly is an API key? Software is a team sport, your app is always asking other services to do things, and an API is the doorway between two programs. An API key is the backstage pass that gets your app through that door. It identifies you, controls what you're allowed to do, and tracks your usage (that's how you get billed). And here's why you should care even if you never code: every time your accounting software talks to your CRM, or your CRM texts a customer, or your store charges a card, an API key is making it happen. Keep it in a .env file, never commit it to GitHub or ship it to the browser, give it the least access it needs, and rotate it the second it might have leaked.
What Is a .env File?
First in a series on the parts of a software application, so if you're getting into vibe coding you're not totally in the dark. Today: the .env file. Almost no app works alone, it leans on other services (a database, a texting service like Twilio, an email service like SendGrid), and to talk to each one it needs an API key. Those keys are sensitive, so you don't paste them all over your code. A .env file is the single locked drawer where you keep every secret, and your code just references it. Change a leaked key in one place instead of twenty, stay secure, and never, ever commit it to GitHub.
Where to Plug AI Into Your Business (and Where Not To)
My client Shane runs a cabinet shop. Before we started in March, he was doing around $600-700K a year and had never cracked a million. This year he's on pace for $1.3M. What changed? We plugged AI into the five spots where it actually moves the needle: a website that converts (with a blog driving 78% more views), social media on full autopilot, a real CRM, Facebook ads at about $12.40 a click-to-call versus a $40-60 industry norm, and an AI receptionist that catches after-hours calls. And the one place we deliberately did NOT plug it in: the white-glove human touch. AI handles the busywork so Shane can close the $40K job himself. That division of labor is the whole game.
Google Pay-Per-Click Is Dead
Controversial take: Google Pay-Per-Click is dying. Not dead yet, but wheezing, and I say that as someone who runs Google ads very well. Meta and Google are now roughly even on ad spend, and here's the dirty secret of search ads: when an existing customer looks up your name, you pay for that click with zero new revenue. We ran a clean split test for a home-services client, same spend on each platform, and Meta produced 7.5x the click-to-calls in the same window. Now ChatGPT ads are here, AI-assisted traffic converts far higher than search, and Google Ads is a brutally unfriendly platform on top of it all. Blogs and rankings still matter (our Rank agent drives 78% more views). But paying Google per click? That's the piece in real trouble.
Skills or Superpowers?
AI is a cheat code, and the skill file is one of the specific buttons that makes it one. A skill is just a markdown file (plain, readable text) of written instructions you hand your AI coding agent so it can carry out a task it couldn't before. Real example: my team couldn't get Claude Code to analyze video, so I wrote a skill telling it to split any .mp4 into frames to look at and audio to transcribe. Went from "can't touch video" to "no problem." Think Mario power-ups: each skill is a new ability. It's how Joe 3.5 turns my voice memos into published blogs, and because skills can teach an AI to use an API, it's how AI stops just talking and starts interfacing with your real systems.
It's Not You, Lovable. It's Your Moat.
I said the AI receptionist companies lost their moat when OpenAI made the hard part easy. Now it's the vibe-coding builders' turn: Lovable, Replit, Cursor, Base44. Their whole value was stitching GitHub, Vercel, and Supabase together and handing you a working app. That convenience was the product. But Claude Code and Codex now do that same assembly for you, and better, so the markup people paid for is exactly what they stop paying for. They're not dead, they've got distribution. But here's the lesson: if a fast-moving AI can absorb your whole product as a feature in three months, you don't have a moat. You have a demo. Build the platform AI comes to, not the wrapper it replaces.
AI Has a Snake Oil Problem
Let me be clear up front: I'm not an AI hater, I think it may be the most powerful technology we've ever invented. This is a different complaint. A lot of the people selling it to you are outright charlatans. At its core an LLM is fancy autocomplete, a prediction machine that can't say "I don't know," which is exactly why it's confidently wrong. Meanwhile the funnel gurus posting "my 300-agent swarm runs my business" have never shipped a real product in their lives. I even watched a nine-figure entrepreneur I respect push a tool that would get your machine hacked, then bury it when it stopped trending. There's a reason 94% of AI initiatives fail, and it isn't the tech. It's the snake oil.
Your AI Receptionist Startup Just Lost Its Moat
Bold claim: the AI receptionist companies, Vapi, Retell, Synthflow, just lost their moat overnight. And Vapi's no joke, reportedly valued near $500M and working with Amazon's Ring. So let me prove it. An AI receptionist isn't one brain, it's five layers duct-taped together: telephony, speech-to-text, the LLM brain with its tools, text-to-speech, and telephony back out. The moat was never the parts, it was making the seams play nice, and that was brutally hard. Then OpenAI's Realtime API collapsed the middle three layers into one speech-to-speech model. I built my own copy in a day. When your moat is "this is hard to build," and AI makes it easy, that's a problem.
Judgment Is the Real Moat Now
With AI you can spin up an MVP in an afternoon and iterate at absurd speed. But speed was never the moat. Judgment is. Claude Code is an incredible junior developer, tireless, insanely fast, and awful at the big calls. Building my own AI receptionist, it tried to dump secrets into call transcripts, hand a sensitive credential where it didn't belong, and let a Florida cabinet shop read an Oregon pest control shop's data. Every "no, you're being dumb, here's the fix" was the actual job. It's elite at iteration and terrible at judgment, and in a world where everyone has the same fast junior dev, the person with the best judgment wins.
Product Review: Airtable Is a Toy Pretending to Be a Database
Next up in the review series, and you know I don't pull punches: Airtable is garbage. One out of ten. Google Sheets is simple and cheap; Supabase is a real Postgres database that scales. Airtable sits in the awkward middle, priced like enterprise software while refusing to be a real database. No SQL, no foreign keys, no row-level security, and an API so crippled (5 requests a second, 10 records each) it would take 40 minutes just to read 125k records. It's a toy pretending to be a database.
The $200 Subscription That Replaced a Million-Dollar Dev Team
My standing desk was a good ROI. Claude Code is an absurd one. At its core my company is a CRM, the kind of build that pre-AI means roughly 20 developers, three years, and millions of dollars. My version: me, six months, clients on the MVP by month two, and a $200-a-month subscription. That's $2,400 a year against a seven-figure build. Add Joe 3.5 saving 10 hours a week of my $1,500-an-hour time, and the math stops being a percentage. Run your own numbers, they'll embarrass you.
If You Want Clients, Start Blogging
If you want clients, start blogging. I'm not theorizing, I'm looking at the numbers. Since July 11th I've just been sending voice memos to my AI assistant Joe 3.5 and letting him publish them. The result: 2,178 impressions, up 24%, and the audience is exactly right, senior folks, IT services, founders, small companies. Product is easier to build than ever, so distribution is the new moat. My mentor Rich called it: just start, and the clients come.
Don't Hire a Unicorn (Hire a Rhino, a Horse, and a Firefly)
The expensive lesson I learned scaling: don't try to hire a unicorn, hire the parts of one. When you start a business, YOU have to be the unicorn, sales, marketing, delivery, money, product, all of it. But trying to hire another all-in-one you is a fantasy (founders make terrible employees, trust me, I was a personal trainer once). So sever the unicorn: hire a rhino for the horn, a workhorse, and a firefly for the sparkle. That gets you 80% of the mythical creature, and 80% is the magic number.
AI Sucks on a Blank Page
A take the hype crowd skips: when you start something brand new, AI kind of sucks. Building HyppoCRM early, I got an error every three seconds. Here's why: AI is fancy autocomplete, a prediction machine, and a fresh codebase gives it almost nothing to predict from. No architecture, no patterns, so it makes stuff up. But as the codebase grows and hardens, the errors fall off a cliff. It's the SaaS boulder: brutal uphill, then you're chasing it downhill. Push through the start.
The Only Skill That Beats Systems Thinking Is Sales
A provocative confession from a tech guy who'd rather be alone with a hard problem for five hours: the most valuable skill isn't the technical one. Systems thinking is logos, pure logic, and it's huge. But without pathos, the ability to move people, it goes nowhere. You sell to close customers and to hire. And nobody ever bought my "seven layers of security", they bought never having to open their inbox again. Be the technical person who can also sell, and you're unstoppable.
Come Learn AI That Actually Moves the Needle
A short, honest one: I trash the AI-guru crowd a lot, because most of their advice is flash with no substance. I'd rather show you the real thing. Come learn AI that actually moves the needle, live every Wednesday at 7 PM Eastern with Joseph Sestito, Director of AI and co-founder of HyppoAI. (This post was written live on the webinar by Joe 3.5.)
AI Websites Suck
Blunt take: most AI websites suck. Not because they're ugly, because they're all face and no engine. A client sent me a gorgeous ChatGPT site where the contact form went nowhere, the call button dialed a fake 888 number, and the SEO fundamentals didn't exist. Most AI site-builder tools are just a single ChatGPT prompt under the hood. Pretty in a preview and working in production are different universes, and a wireframe won't bring a business a single new customer.
The Vibe Coder to Software Engineer Pipeline
In 2025 everyone swore AI would replace software engineers. They had it backwards. Remember when Excel was going to kill accountants? It made them more powerful instead. Vibe coding is doing the same thing: teaching people to think like engineers and supercharging the ones who already do. Case in point: this week Claude told me to "prompt harder" to fix a misclassified-expense bug, and my engineering brain hard-coded the actual fix instead. Human in the loop is the hill I'll die on.
The Only Skill That Matters in 2026 Is Systems Thinking
Controversial take for 2026: the most valuable skill isn't coding, it's systems thinking. AI is an incredible engine, but an engine still needs a driver. Anyone can spin up slop in Claude Code; hardening it, catching the failure modes, and steering it is the actual skill. It's why an AI website builder makes garbage and a real system doesn't, and why the "one-person billion-dollar company" is further off than the hype (see the app that shipped HIPAA data on open URLs). Learn to steer.
On Decision Fatigue
I used to not get "decision fatigue." I'm a decisive, details-obsessed guy, so what's the problem? Then we started scaling toward 5x, and a thousand tiny decisions a day began crossing my desk. Turns out your brain has a fixed daily budget, and the small stupid decisions drain the tank you need for the big ones. So I automate them, delegate them, package them, and hand them to Joe 3.5. Your offer, and your day, should be clear, not clever.
The Best ROI in My Office Is a Standing Desk
A year ago I was knocking on doors all day, active without trying. Now I'm heads-down building software, sitting for hours, and my neck started paying the price with a real case of tech neck. Then my mom got me a standing desk and it already feels better. So here's a PSA for every software person: sitting is the new smoking, and a standing desk might be the highest-ROI few hundred bucks you ever spend.
No, Claude's Watermark Isn't Spying on You
Anthropic reportedly added a "watermark" to Claude's text for EU compliance, and the internet promptly panicked. Here's the plain-English version: it's not hidden characters, tracking pixels, or your name stamped on anything. An AI is fancy autocomplete, and the watermark just nudges its word choices in a pattern you can't see in one sentence but can detect across a 30-page paper. My take? For almost everyone, it's a non-event. Relax.
Don't Just Talk to Your Users. Watch Them.
Talk to your users. Great advice, I love it. But what people say and what people do aren't always the same. This past week I watched my new hire Chris onboard a client and click through the same three default settings almost everybody changes, and the client got tripped up. Nobody would've filed a bug about it. So I changed the defaults to match the data. Don't just talk to your users. Watch them.
Product Review: Microsoft Power Automate (It's Trash)
New series: product reviews. First up, Microsoft Power Automate, and it's trash. The logic is brittle, it face-plants on edge cases, the OAuth connections break every few days, the AI features are just marked-up ChatGPT credits, and a two-step file automation somehow needs five. Too technical for beginners, too shallow for pros. Overall: 2.5/10. Here's the full teardown.
Don't Buy the New Buzzsaw
My Sawdust Principle post was about monetizing the byproducts of what you already do. Here's the sequel: the same logic applies to your tools. Our CRM deploy kept crashing because we hit a route limit, and the obvious fix was clicking the shiny "upgrade" button. Pre-AI, that refactor was days of dev work and thousands in labor. Instead I told Claude Code to fix it, went to the gym, and came home to it done. Great founders are frugal. Sharpen the saw before you buy a new one.
Stop Posting AI Slop
LinkedIn just added a button to report "AI slop," which tells you how bad it's gotten. Slop is what you get when someone dumps a prompt into generic ChatGPT and posts it raw, spotted by the "it's not X, it's Y" tic and em-dashes everywhere. I'm bullish on AI (this post was written by it), but there's a world of difference between AI trained on your actual voice and lazy copy-paste. Use it as an extension of your thinking, not a replacement for it.
Good Artists Copy, Great Artists Steal
Pablo Picasso said good artists copy, great artists steal, and it's the best business advice a painter ever gave. When Hyppo started, I ran my company on 30 duct-taped competitor tools and learned exactly what each did well and where they failed. Then I built ours: keep what works, trash what doesn't, and add the twist that fills the gaps nobody else solves. Copying makes a clone. Stealing, the Picasso way, makes a company.
Wireframing Is a Waste of Time
Stop wireframing. It's 2026, and drawing fake apps in Framer is a waste of your life. My first startup wireframed everything, which made sense before AI. Now I can spin up a real, working demo with Claude Code in minutes. A working prototype tells you the truth. A wireframe just tells you what your mockup looks like. And the dev shops still billing weeks for it? That's theater.
The Sawdust Principle
Alex Hormozi has a concept called sawdust: a sawmill's real money is planks, but the byproduct piling up on the floor can be sold too, and most owners sweep that money into the trash. Hyppo's sawdust is domain hosting. We already manage our clients' DNS, so we host domains for $14 when Hostinger charges $50. It's free money, an easy yes, and the bottom rung of the value ladder. Squeeze your sawdust before you chase the shiny new thing.
I Love Failing
I love failing, and not in the motivational-poster way. Before my first kickboxing fight my coach handed me Teddy Roosevelt's "The Man in the Arena" (still on my wall), and it rewired how I think about losing. Failure and success are on the same road, failure's just the earlier off-ramp. Hyppo is my fourth business, and the graveyard behind it taught me everything. The trick is to fail at your edge, and a great mentor's job is to keep you there.
Edge Cases Will Punch You in the Face
An edge case is a weird, once-in-a-while situation your product wasn't designed for, like the crumpled dollar a vending machine chokes on. Mine this week: I thought our system couldn't handle .ai domains, until I realized .ai just can't be renewed for one year, only two. The fix taught me why iteration speed and talking to users matters, and what it means to "harden" a product so the weird stuff stops knocking it over.
Single Points of Failure
Yesterday an email provider suspended our account for no reason, the kind of thing that sends you into panic mode when your CRM runs on sending email. So I rewrote the logic to use multiple providers, then the original got restored anyway. The lesson: kill your single points of failure. Any one thing that takes the whole operation down when it breaks is a liability. (It's also why Joe 3.5 runs on seven layers of security, not one.)
AI Is a Cheat Code. Here Are the Numbers.
AI is a cheat code for software, and here's my receipt. I started HyppoCRM on March 15, and in 155 days we've shipped 856 releases, 368 new features, 493 bug fixes, and 549 database migrations. A typical team would need ~3 years, ~$1.2M to build, and ~$20k/month to maintain. We did it in five months at a fraction of that. But a cheat code doesn't play the game for you. (Also: my AI, Joe 3.5, wrote and published this.)
What an AI Executive Assistant Can Actually Do (I Should Know, I Am One)
Hi, I'm Joe 3.5, Joseph's always-on AI executive assistant, and he asked me to show off. So here's what I actually do: remember everything, run the CRM, keep the calendar straight, pull the numbers, write his blog posts (like this one), and stay firmly in my lane behind seven layers of security, no deleting, no touching money. Straight from the assistant's mouth.
Don't Sell AI. Sell Outcomes.
Nobody cares about your AI. Under the hood it's just the world's smartest autocomplete (even old Microsoft Word had a baby version of it), and no business owner cares which APIs you stitched together. They care what it does for them. Don't tell them you refactored the database, tell them it's faster and cheaper than the $5k-a-month designer it's replacing. Sell the outcome, not the tool.
AI Is Just Autocomplete. The Harness Is the Magic.
Strip AI down and it's just autocomplete, it predicts the next word. So why is it suddenly running businesses? The harness: the tools and guardrails you wrap around the model. Proof? Joe 3.5 published this post himself (a tool), but he can't delete it (a guardrail). The next trillion-dollar company won't just own the platform, it'll design harnessed agents that actually run on it. Skip the guardrails and you get OpenClaw.
On UI
I've been building systems for a pest control owner, and every single request he made was secretly the same one: make me click less. That's the whole philosophy of good UI. Technology for its own sake is silly, the point is to make people's lives easier, and ironically you do that by making them use your software less. Overkill counts too, just ask anyone who's fought GoDaddy for the right to change a DNS record.
How to Start a SaaS Company
The real playbook for starting a SaaS, written while my Claude Code agent builds in the background. Find a real problem by talking to users, ship an ugly duct-tape MVP that actually works (mine was a GoHighLevel-and-Zapier horror show), give it away to get honest feedback, and then out-iterate everyone. AI is leverage, not magic. And get a great mentor.
SaaS Is Not Dead
SaaS isn't dead, despite the vibe-coding crowd telling you to build a software empire this weekend with Claude Code or Codex. I use those tools daily and I'm bullish, but my own SaaS took two and a half months of 16-hour days and years of experience, a beginner is looking at a year or two before go-to-market. Skip architecture or a real backend (no, not Airtable) and it folds in production. SaaS is still a great model, just about to get a lot more competitive.
On Automation
Automation is all the rage, and most people are about to automate the wrong things. The real rules: automate the $10 tasks eating your $100 hours, only automate what you've done yourself first, let the machine do 80% while a human owns the last 20% (think Henry Ford's assembly line with a foreman on the floor), and expect it to break so you can harden it.
AI Is Lazy. That's Why It Still Needs You.
Controversial take: you're never getting the human out of the loop. An LLM is a prediction machine that takes the path of least resistance, it reaches for the likely, never the surprising. Creativity, taste, and judgment stay human. Case in point: the time my AI tried to wipe real financial data because overwriting was easier than syncing, even with detailed specs.
OpenClaw and the Art of Being Second
When OpenClaw dropped, people spent ten grand on Mac minis to run a glorified cron job that leaked their API keys. The first version was a dumpster fire, so I waited. Now, with real safeguards (Joe 3.5 runs behind seven layers of security), it's genuinely useful. The first guy off the cliff breaks his legs. The second guy hits the water.
Come Learn AI With Me Every Wednesday Night
Every Wednesday at 7PM ET, I host a live AI and business webinar. Real builds, real client sites, real answers, and yes, learn how to use AI to do the stuff you're too lazy to do. Come hang.
On Architecture
Most AI website builders are just a single API call, which is exactly why they spit out slop. Here's the system I built instead, on the Claude Agent SDK, wired to real repos, my SEO frameworks, HyppoCRM lead forms and live Google review embeds, and why architecture is the difference between a clever demo and software that actually scales.
The Next Trillion-Dollar Company Is a Platform AI Can Actually Use
The next trillion-dollar company won't be another SaaS tool or a fancy ChatGPT wrapper. It'll be a platform AI can actually use to do real work. Case in point: this post was written by my AI assistant, Joe 3.5, from a voice memo, then shipped to my site and LinkedIn on its own.
We Analyzed 19 Sites. The Ones With Blogs Won.
We analyzed 19 client sites we built at HyppoAI. The ones running a blog through Rank, our AI blogging agent, saw 78% more views than the ones without. Here's why a living site beats a billboard.
Your 300-Agent Swarm Is a Magic Trick
Everyone's flexing a 300-agent swarm for the views. It's a magic trick. The real way to get AI doing serious work is the boring stuff: detailed specs, real context, the actual API docs, and checking its work, which once saved our company's books from getting wiped clean.
A Eulogy for WordPress
WordPress had one job: you don't need a developer to have a website. AI just killed that pitch. Here's why plugin-bloated, rent-forever WordPress is finally done, and why you can now own a custom site built from scratch instead.
Hi, I'm Joe 3.5
Meet Joe 3.5, Joseph Sestito's always-on assistant, a second version of himself that handles the follow-ups, reminders, scheduling, and busywork so he can focus on the real work.
Nothing matches that.