Over the past two weeks, I've launched a live product, written 160 blog posts, and started a second project — all using AI as my co-founder. This page explains what I'm doing, why it matters, and where it's going.
I'm building software products — the kind you'd normally need a team of 5-10 people to create — except I'm doing it alone, with AI as my engineering partner, marketing team, and business strategist. All from a single tool called Claude Code.
Claude Code is an AI tool made by Anthropic (the company behind Claude). It lives in my terminal — the text-based interface developers use to talk to their computers. I describe what I want in plain English, and it writes the code, creates the content, runs the tests, and deploys the result. It's like having a very talented employee who works at the speed of thought and never sleeps.
The problem: When you ask ChatGPT or Claude to write something, it sounds generic. It doesn't sound like you. MyWritingTwin solves this by creating a detailed “AI Writing Profile” — a document that captures your unique voice, style, and preferences, so AI tools can write exactly the way you do.
You answer questions about how you communicate and provide a few writing samples. Our AI analyzes your style deeply — sentence patterns, vocabulary choices, tone, even your quirks — and produces a comprehensive instruction document. You paste it into ChatGPT, Claude, or Gemini, and suddenly the AI writes like you.
Professionals who use AI daily but are frustrated that it sounds robotic. Bilingual executives who switch between languages. Content creators who need consistency. Consultants who write 50 emails a day. Anyone who thinks “AI is useful but it doesn't sound like me.”
The user dashboard showing the 4-step Writing Twin journey.
Product explainer video — animated scenes produced by Remotion, narrated by an AI avatar via HeyGen. Also generated from Claude Code.
MyWritingTwin.com is a real, working product accepting payments in USD, EUR, and JPY. It has user accounts, a dashboard, Stripe payments, AI-powered writing style analysis, audio transcription, email sequences, and multilingual support in English, Japanese, French, and Spanish. Officially launched February 14th, 2026 — built on two weekends and a few evenings.
The problem: Presentations are full of static diagrams that fail to communicate complex ideas. Animating them requires expensive motion design tools and skills most people don't have. FluxDiagram lets you describe what you want to animate in plain English, and it generates professional animated diagrams you can embed directly in your slides.
You type “Show data flowing from the user's browser through our API gateway, hitting the cache first, then the database, with the response flowing back” — and FluxDiagram produces a smooth, professional animation you can drop into PowerPoint, Google Slides, or Keynote. No design skills needed.
Currently in the strategy and architecture phase. The business plan, pricing model, brand voice, go-to-market strategy, and technical research are complete. The cost model has been validated across 6 AI models. Development begins soon. Target: a $14-29/month SaaS with a free tier.
FluxDiagram is being built using the exact same tools, skills, and methodology as MyWritingTwin. The Claude Code operational system I created for the first product is being reused for the second. This is the whole point — the system scales.
Each product targets a different problem, a different audience, and a different market — but they all share the same operational backbone.
Creates personal AI Writing Profiles for professionals who want AI to match their unique communication style.
Generates professional animated visuals for presentations. Describe what you want, embed the result in your slides.
A third product is in the conceptual phase. The playbook is proven, the tools are ready, and the process is repeatable.
Here's what most “I built this with AI” stories leave out: you can't just have an idea and ask ChatGPT to build you a business. AI is the engine, but the driver needs to know where they're going. What makes this work isn't just the AI — it's three decades of expertise compressed through AI execution.
30 years of building software — entirely self-taught, starting as a hobby and now turning it into a real business. Decades of hands-on experience with everything that makes systems like this work: architecture, databases, APIs, security, deployment. Not writing code anymore — directing AI that writes code. The experience validates what the AI generates.
Years of building and selling software products. Understanding pricing psychology, conversion funnels, SEO strategy, content marketing, email sequences. The AI executes the playbook; the experience wrote it.
Deep understanding of how large language models work — their strengths and limitations. Knowing how to prompt effectively, when to trust AI output and when to push back, how to build systems so AI can operate reliably at scale.
The individual products are interesting. But the truly exciting part is what sits underneath them: a reusable system for building and operating entire businesses with AI. We call it agentic engineering — the human directs, the AI executes.
Think of it like a restaurant franchise. The first restaurant is hard — you figure out the recipes, the supply chain, the training manuals, the operations playbook. The second restaurant is easier because you have the playbook. The third is easier still. Eventually, the playbook itself is more valuable than any single restaurant.
Each layer amplifies the one below it. The same stack is now building FluxDiagram.
Development & QA: Writes code, runs tests, manages the database, deploys to production. The Quality Gate agent runs 7 automated checks before every deploy — TypeScript, tests, build, translations, links, SEO, and secret scanning.
Marketing & SEO: The Content Pipeline agent produces end-to-end blog posts with images, infographics, and social drafts. The SEO Monitor tracks rankings, detects drops, and finds content gaps. The AEO Infrastructure agent ensures AI crawlers can cite the site. All across 4 languages, 160+ posts, and growing.
Operations & Analytics: The Daily Briefing agent synthesizes PostHog, GSC, Stripe, and API cost data into actionable reports. The User Lifecycle agent monitors customer journeys, detects stuck users, and drafts interventions. The Product Bootstrap agent scaffolds entirely new SaaS products from a proven template.
Traditionally, building and running a software business requires at minimum: a developer, a designer, a content writer, a marketing person, and a business strategist. That's 5 salaries, coordination overhead, and communication friction.
This system compresses all of those roles into one person directing AI. The fixed cost? About $5/month in domain names. Everything else runs on free tiers. Profitable from the first customer.
These aren't projections. This is what has actually been built, committed, and deployed since January 31st, 2026 — on weekends and evenings, for MyWritingTwin alone.
All of this was built in roughly 70 hours of actual work across 72 sessions — about 28 hours on weekends, 20 hours on weekday evenings, spread over two weeks around a full-time job.
A traditional startup team (developer, designer, content writer, marketer, strategist) would need an estimated 1,680 hours to produce equivalent output. That's a 24x efficiency multiplier — not from working harder, but from directing AI to execute what experience has already mapped out.
All of this was built on weekends and evenings — nights after the day job, Saturday mornings, Sunday deep-work sessions. No sabbatical, no time off. Just focused side-project hours and an AI that doesn't sleep.
Started Saturday morning with an idea: what if AI could learn to write exactly like a specific person? By Sunday night, the project structure, database, authentication, questionnaire system, and AI voice analysis engine were functional. 100+ commits across two full days.
After-hours sessions: integrated Stripe payments in 3 currencies, built the user dashboard, started the blog system, and wrote the first batch of SEO-optimized posts. 2-3 hour blocks each night, compounding fast.
Full weekend dedicated to content and go-to-market. Created the content pipeline skills, generated images, wrote 30+ blog posts, built the brand voice guide. The system was proving it could produce marketing assets at scale.
Started strategic planning for the second product in evening sessions. Fixed critical Stripe checkout bugs, added upgrade paths, built video production components, expanded to 160 blog posts across 4 languages. Conceived FluxDiagram and wrote its full business plan.
Generated 39 AI infographics, translated every blog post into Japanese, French, and Spanish, built the Claude Code operations presentation. Some of this time was also spent waiting for Stripe to approve the payment integration.
MyWritingTwin.com goes live — a fully operational SaaS with payments, AI generation, multilingual content, and a full marketing engine. Built on weekends and evenings by one person with an AI co-founder.
Each product solves a real problem for real people. MyWritingTwin helps professionals communicate better with AI. FluxDiagram helps presenters explain complex ideas visually. These aren't toy projects — they're designed to generate revenue and grow.
This is an early proof of what “AI-augmented entrepreneurship” looks like. One person, directing AI, building things that used to require teams. The tools will only get better. The cost will only go down. The speed will only increase. We're at the very beginning.
Traditional SaaS companies raise millions in venture capital to pay engineers, designers, and marketers before earning a single dollar. This operation runs on free tiers for hosting, database, email, and analytics. The only fixed cost is domain names — about $5/month. Variable costs kick in only when someone buys: ~$1.50 for the AI API call that generates their profile, plus ~$4 in Stripe transaction fees. Total: about $5.50 per customer, and zero otherwise.
The biggest line item? Claude Code at $200/month — the AI co-founder that handles engineering, content, marketing, and operations. For context, a single junior developer costs $5,000+/month. The entire investment to build and operate a SaaS business is less than a team lunch.
Profitable from the first customer. No investors, no burn rate, no runway countdown. As the business scales, some services will need paid plans, but the point is: you can validate a business for the price of a domain name and an AI subscription.
There's something recursive happening here. MyWritingTwin teaches AI to replicate a person's writing style. But the business itself was built by directing AI that operates like an ideal engineering team — writing code, generating content, managing deployments, running analytics. The process is a live demonstration of the product's core thesis: AI can be made to work the way its operator works.
The system that built MyWritingTwin — the playbooks, guardrails, automations, and operational workflows — isn't just tooling. It's a reusable engine for launching products. The first product took two weeks. The second should take less — though product complexity varies, the compounding effect of the system is real. Each iteration adds reusable patterns and operational muscle. That's the meta-product: not a single SaaS, but a machine for building SaaS businesses.
And this page? It's the proof. “A SaaS built in 14 days using Claude Code” isn't just a case study — it's a demonstration of exactly the kind of AI capability that MyWritingTwin is selling. The story markets the product. The product validates the story. The loop closes.
The system is built. The first product is live. The second is loading. Here's the roadmap.
The product works. The content is driving organic traffic. The immediate focus is converting visitors into paying customers — optimizing the funnel, fixing friction points, and building the email sequences that nurture leads.
With the strategic planning complete, it's time to build the second product. Thanks to the reusable system, this should be significantly faster than the first. The architecture decisions, payment integration patterns, and deployment pipeline are already proven.
After two products are running on the same system, formalize it into a reusable project template. The skills, hooks, scripts, and operational workflows become a starter kit for future products — or potentially a product themselves.
The endgame is a portfolio of small, profitable SaaS products — each serving a specific niche, each generating recurring revenue, all operated from the same terminal by the same system. Not one big bet, but many small, smart ones.