Timestamped February 14, 2026. Written down May 26, 2026.
In February, I went on record with a prediction. I told my team that by June 2026, every major website platform — Wix, Squarespace, Duda, Webflow, Shopify, WordPress — would have some form of talk-to-build. Conversational AI where you describe what you want, and the platform builds it. No dragging. No dropping. No code. You talk, and it builds.
Three and a half months later, it’s happening faster than I called it.
WordPress 7.0 shipped the MCP Adapter — any AI assistant can read and edit your site directly. Webflow launched an official MCP server that connects to Claude. Duda released a full AI stack. Wix Harmony is doing the same. Shopify shipped Sidekick. Squarespace AI is live.
The talk-to-build era didn’t arrive in June. It arrived this spring, and the platforms are still racing each other to ship more of it.
So here’s the timestamp on the actual manifesto: the people who learn to direct AI right now, this year, are the ones who get to build the next five years. Everyone else is going to spend that time watching their businesses get smaller.
What I actually believe
I believe in a 90/10 split between AI and humans.
AI does the 90% that used to take you all week — drafting the copy, building the page, generating the imagery, wiring the automations, running the QA checks, writing the reports. Humans do the 10% that AI can’t touch — the taste, the strategy, the relationship, the creative call that breaks the pattern, the judgment about what to build in the first place.
That 10% is where the new careers live. It’s where the money is. It’s where the meaning is. The people who figure out how to run that split well are going to build companies the old playbook couldn’t dream of. They’ll do it with smaller teams, lower costs, and outputs that look like they came from 50-person agencies.
But the split only works if you commit to both sides. If you skip the AI 90%, you’re stuck in the old economics. If you skip the human 10%, you’re shipping the same generic AI slop everyone else is. The harmony is the whole point.
What is AI-First?
AI-first means you stop asking “should we use AI for this?” and start asking “where is AI NOT doing this yet?”
Most companies still treat AI like a tool they sometimes use. AI-first treats AI like the default operating system — every workflow, every deliverable, every internal process starts inside an AI loop. The human comes in for taste, judgment, and the calls AI shouldn’t be making alone. Everything else gets handed to the machine.
Here’s how it shows up in practice. I’ll use my own industry.
Building a website used to take 45 to 90 days. Some agencies still book six months. Discovery calls, wireframes, design reviews, dev sprints, QA, launch, three rounds of revisions, more launch dates.
Now I build the same website in four days. Not forty. Four.
Same outcome. Same quality. Less than a tenth of the time.
All I did was remove a zero.
┌─────────────┐ ┌─────────────┐ │ 40 days │ remove → │ 4 days │ │ (old way) │ the zero │ (AI-first) │ └─────────────┘ └─────────────┘
You don’t need to tell clients how you operate. You don’t need to explain Claude Code, MCP servers, agentic workflows, or any of the plumbing. They don’t care about the plumbing.
What they care about is one thing: speed is the currency.
A business owner who wants a new site doesn’t want to wait three months. A founder launching a product can’t sit through a 12-week sprint. A CEO who needs a landing page yesterday isn’t going to be patient. The moment you can say “four days instead of forty,” you’ve already won the conversation. The “how” doesn’t matter — the speed does.
That’s AI-first. The plumbing is invisible. The speed is the brand.
Why this is happening now
This is a disruption that rhymes with the last big one.
In the 1910s and 1920s, farms ran on horse and carriage. Everyone knew the routine. Some farmers couldn’t even afford a horse, but they knew how the world worked. Then in the 1950s, tractors got cheap enough to spread. The naysayers said tractors were sloppy. They said the engine would break, the parts would cost too much, the farmer would lose his connection to the land. None of that mattered. Within twenty years, you either had a tractor or you went out of business.
AI is the tractor. Everybody saying it’s sloppy, it’s not the future, it’ll take all our jobs — they’re the people who picked the horse in 1955. The choice is the same. Automate or die.
A real example
I’ll give you a real example so this isn’t abstract.
There’s a guy in California — let’s call him Frank. He’s 65 years old. He’s been running an entertainment news blog for the last fifteen years. He used to get 30,000 visitors a day. Last year, that site was generating about $125,000 a year. This year, he’s making about $30 a day. Same content. Same domain. Same effort.
What changed? People stopped browsing websites the way they used to. They started asking Google’s AI mode. They started asking ChatGPT. The aggregators pull from the sites that have schema, citations, llms.txt files, and structured FAQ blocks. The ones that don’t have those are invisible. Frank’s site is invisible.
He’s doing nothing about it. That’s the part that kills me. The platform shifted, and he didn’t shift with it.
A lot of businesses are about to be Frank. WordPress sites with 70,000 monthly visitors are dropping to zero, not because they got worse, but because the world stopped looking at them the same way. The browsing habits changed. The discovery model changed. The traffic that used to come from Google searches now goes to Google AI Mode, ChatGPT, Claude, Perplexity. If you’re not cited there, you don’t exist there.
The website still matters. The role of the website just changed. It used to be a destination. Now it’s a source the AIs cite. If your site isn’t built for that, you’re done.
Who this manifesto is for
I’m writing this for a specific group of people.
Business owners, founders, CEOs, and VPs of marketing running real companies, somewhere between $1M and $100M in revenue. You’ve built something. It’s working. But the ground is shifting under it, and you can feel it. You don’t know yet what to do about it. You want a clear path that doesn’t require you to learn to code or hire a 20-person AI team.
Software engineering students graduating right now, or developers a few years in, looking at the layoffs and wondering what just happened. The jobs that used to exist are getting cut by the thousand. The jobs that are replacing them look completely different — you don’t write code anymore, you orchestrate AI that writes code. You work in Claude Code or Google Antigravity. You direct, you don’t type. If you don’t pivot fast, the door closes.
People in mid-career switches. You hate what you do. You’ve been thinking about leaving. The AI shift is the cover story you’ve been waiting for. The career you want next probably didn’t exist 18 months ago, and that’s not a bug, it’s the feature.
Designers, creative directors, agency owners. You know your work is good. You also know your hourly rate has a ceiling. The combination of taste plus AI automation breaks the ceiling. Most of your peers won’t learn it in time. The ones who do are about to print money for the next five years.
D2C brands and the people running them. Your customers are now researching with AI, shopping with AI, and sometimes letting AI buy for them. Stripe ACP and Shopify Agentic Storefronts are already live. If your store isn’t ready for agents to transact, you’ll lose share to the stores that are.
If you’re in any of those groups, this is for you.
What I’m offering
I’m here to be the guide.
I want to be Gandalf to your Bilbo. Morpheus to your Neo. Yoda to your Luke. Pick your metaphor. The point is the same — you have a fight ahead of you, you don’t yet know the rules of it, and I’ve been in the trenches long enough to walk you in.
Here’s why I get to say that. I’m the founder and CEO of MK-Way — a seven-figure design agency I’ve been running for six years. Seven years total as an entrepreneur. I’ve shipped websites for clients on Shopify, WordPress, Webflow, and Wix the entire time. The agency works. The team is real. The clients are real.
And I’m now pivoting the whole company to AI-first. Every system, every service, every deliverable. I’m doing it in public — on mikekwal.com, on Talk-to-Build, on YouTube, on Instagram, on LinkedIn. You’re going to see what works, what doesn’t, and what I’d do differently. I’m not selling you a course on what I think AI might do. I’m shipping the pivot live and telling you what’s happening in real time.
That’s what this manifesto is. An invitation to follow along while I do this in public, and to build alongside me if you want to.
What we’re going to build together
Real things. Not theory.
- Your first AI-native website
- Your first Shopify store with agentic commerce wired in
- Your first AI agent that runs a workflow you currently do by hand
- Your first program — a tool you own that solves one specific business problem
- Your first operating system for your own business — the loop that runs without you
- An app, if that’s what you need
All of it inside Claude Code and Google Antigravity. Not because those are the only tools, but because those are the cheapest, fastest, and most powerful way to ship right now. I’ll show you the difference between agents and LLMs — the LLM is the brain, the agent is the worker. You need both, and you need to know which is doing which.
I’ll show you how to think about cost. The work an employee used to do for $4,000 a month, you can now do for $0.04 a day in AI tokens, maybe $40 a month if you run a lot. I built the cron that generated yesterday’s blog post on my site for less than a dollar a month in API costs. That’s not magic, it’s just the new math.
The mission
I’m on a mission to help 100,000 people make the shift to AI-First.
Some are business owners who want to keep their business alive through the transition. Some are graduates whose first job doesn’t exist anymore and need a new path. Some are mid-career people who’ve been waiting for a reason to switch. Some are agency owners and designers who want to ride the wave instead of being washed out by it.
I’m going to teach what I’m actually shipping. I’ll write about the wins and the breaks. I’ll publish the specs people can download and run themselves. I’ll keep growing MK-Way as the live test case, so you can see the playbook running on a real company.
The point of this manifesto isn’t to predict the future. The prediction is already happening. The point is to invite you to build the next part of it with me — instead of watching it go past you.
If you can talk it, you can build it.
— Mike Kwal
Founder + CEO, MK-Way