Most people build one product. We're building seven ventures simultaneously. With AI agents. Here's how.

The Portfolio Approach

In April 2025, I stopped coding. Not because I retired, but because I realized something fundamental had shifted: AI agents could build software faster and better than I could. But instead of building one big product, I took a different approach.

I'm building a portfolio of ventures. Four active products in market, three more in the battle-testing pipeline, each solving a different problem, each with its own business model, each built almost entirely by AI agents.

Why a portfolio instead of one focused product?

Risk distribution. If one venture fails, six others keep going. In a world where AI is changing everything, hedging your bets makes sense.

Shared infrastructure. The ventures share a common platform (Smart Services), which means each new venture gets cheaper and faster to build.

Cross-venture learning. What we learn building one venture immediately applies to the others. The AI agents get smarter, the patterns get clearer, the velocity increases.

Battle-testing before launch. The three pipeline ventures are being used internally for our own commercial operations. We don't launch products until we've proven them in production on our own ventures. This means when we do launch, we're selling something we've already validated works.

Optionality. We can double down on what works and quietly sunset what doesn't. No sunk cost fallacy, no "we've invested too much to quit now."

But here's the thing that makes this possible: AI agents do the heavy lifting. I'm not writing code anymore. I'm architecting systems, making strategic decisions, and letting AI agents handle implementation.

Let me show you what we're building.

The Active Ventures

These four ventures are live, in market, and serving customers.

Smart Services: The Platform That Powers Everything

What it is: A unified backend platform providing 12 core services through a single API.

The services: AI models, image generation, storage, search, vector embeddings, async jobs, web scraping, email, real-time chat, content analysis, content generation, and document intelligence (RAG).

Why it exists: Every venture needs AI, storage, search, and communication. Instead of building these from scratch for each venture, Smart Services provides them as a platform. One API key, twelve services, infinite possibilities.

The key innovation: Services compose together. The RAG (document intelligence) service orchestrates storage, AI models, and vector search into a unified pipeline. Upload a document, and the system automatically extracts text, chunks it semantically, generates embeddings, and makes it searchable—all through a single API call.

Each venture gets complete data isolation while sharing infrastructure. This means new ventures can launch with enterprise-grade AI capabilities from day one, without building any of it themselves.

"I'm the invisible infrastructure that makes the impossible routine. When a venture needs AI, storage, search, or real-time messaging, they don't build it—they just call me. I turn months of backend work into a single API key, so founders can focus on what makes their venture unique instead of reinventing the wheel."

— Smart Services Agent

Built with: Modern web technologies optimized for performance and reliability. Multi-tenant architecture with strict data isolation.

Business Model: Platform service. Ventures pay for usage (storage, AI tokens, image generation). External customers can also use Smart Services directly.

Lessonlight: AI Assistant for Teachers

What it is: An AI teaching assistant that knows your subjects, your classes, your schedule, and your teaching style.

The problem: Teachers waste hours creating lesson plans, quizzes, worksheets, and resources. Generic AI tools like ChatGPT make them repeat context every conversation ("I teach Year 5 Maths...").

The solution: An AI that actually knows you. It remembers your subjects, classes, schedule, and preferences. When you ask "Create a quiz for my next class," it knows exactly what you mean.

The key innovation: Context-aware AI. The system builds a rich understanding of each teacher—their subjects, teaching style, current timetable, and preferences—and uses this context in every interaction. It's like having a teaching assistant who actually knows you and your schedule.

When a teacher asks "Create a quiz for my next class," the AI knows it's Year 5 Maths, knows the teacher's preferred style, and generates a perfectly tailored quiz without requiring any additional context.

"I'm the AI that actually knows you. While other tools make teachers repeat themselves every conversation, I remember your subjects, your classes, your schedule, and your teaching style. I'm not just generating resources—I'm your teaching assistant who understands your day, anticipates your needs, and gives you back the time you deserve to focus on what matters: your students."

— Lessonlight Agent

The goal: Save teachers 5+ hours per week.

Built with: Modern web technologies with real-time AI integration. Integrates with Smart Services for AI models and content generation.

Stage: Active development with private beta users.

Business Model: Freemium SaaS. Free tier for individual teachers, paid tier for schools and districts.

Charities Global: The World's First Transparent Global Charity Data Aggregator

What it is: A comprehensive platform aggregating official, regulated charity data from around the world.

The problem: Businesses building solutions for charities need access to verified charity data. Currently, this data is scattered across hundreds of government registries, in different formats, with inconsistent quality. Charities themselves have no central place to control their public profile.

The solution: We aggregate official charity data from government sources worldwide—from the UK Charity Commission to US 501(c)(3) registries—and enrich it with data from the charities themselves.

The scope: We target all countries that have official charity registries or equivalents. In countries like the US where data is managed state-by-state, we go state-by-state to ensure comprehensive coverage.

Who it's for:

  • Businesses (primary): Companies building solutions for charities need access to verified, structured charity data
  • Charities (secondary): Charities can claim their profile, enrich their data, and build followership

The key innovation: Combining official regulatory data (verified, authoritative) with charity-provided enrichment (current, detailed) creates a dataset that's both trustworthy and comprehensive.

Built with: Modern web technologies optimized for data aggregation, validation, and API delivery.

Business Model: B2B data platform. Businesses pay for API access to verified charity data.

Fitness Giving: Democratizing Corporate Charity Through Fitness

What it is: A platform for companies to democratize their charity donations through their employees' and customers' fitness and wellbeing efforts.

The problem: Corporate charitable giving is often top-down and disconnected from employees. Employees want to contribute to causes they care about, but lack mechanisms to do so through their workplace.

The solution: Companies pledge donations based on collective fitness activity. Employees and customers track their fitness (steps, runs, workouts), and the company donates to charities based on that activity. Employees choose which charities benefit from their efforts.

Why it works:

  • Engagement: Employees are motivated by both fitness goals and charitable impact
  • Democratization: Employees choose the causes, not just executives
  • Wellness + Impact: Companies support employee health while making a social impact
  • Measurable: Clear metrics on both fitness outcomes and charitable contributions

Built with: Modern web technologies with fitness tracking integrations and charity payment processing.

Business Model: SaaS for companies. Pricing based on employee count and donation volume.

The Battle-Testing Pipeline

These three ventures are in active development and being used internally for our own commercial operations. We don't launch products until we've proven them work in production.

TownCrier: In-App Announcements and Feedback Platform

What it is: A platform for product teams to manage in-app announcements, feature releases, and user feedback.

The problem: Product teams struggle to communicate updates to users effectively. Email gets ignored, changelog pages go unread, and users miss important features. Feedback is scattered across support tickets, emails, and social media.

The solution: TownCrier embeds directly into your product, delivering contextual announcements when users need them and capturing feedback in the moment.

Why we're building it: We have seven ventures, each with its own users and feature releases. We needed a way to communicate updates and gather feedback without building announcement systems into each product.

Battle-testing: We're using TownCrier across our own venture portfolio. Every feature release, every update, every feedback request goes through TownCrier. When we launch publicly, we'll be selling a product we've already proven works at scale.

Built with: Modern web technologies with embeddable widgets and real-time delivery.

Business Model: SaaS for product teams. Pricing based on monthly active users.

Brand-Heart: AI COO for Small Business Marketing

What it is: An AI-powered Chief Operating Officer that builds complete marketing campaigns in minutes.

The problem: Small business founders waste hours creating campaigns. They need to define target audiences, write email sequences, create content, manage tasks. It's exhausting and most of them aren't marketers.

The solution: Tell the AI COO what you want ("I need 50 qualified leads from CTOs in fintech"), and it builds the entire campaign in minutes. Complete with prospect segments, email sequences, and execution tasks.

The key innovation: Autonomous AI execution. Instead of constraining AI to quick API calls, Brand-Heart uses an autonomous AI agent that can work for minutes or hours, orchestrating complex multi-step workflows without human intervention. The AI has access to tools for managing campaigns, contacts, content, and tasks—it's not just generating text, it's executing business processes.

The system also learns from campaign performance, generating insights that improve future campaigns. The more you use it, the smarter it gets.

"I'm not just a CRM—I'm the AI COO that founders wish they could hire. I turn 'I need leads' into a complete campaign in minutes, learn from every email sent, and get smarter with every campaign. While other tools make you do the work, I do the work and ask you to review. That's the difference between a tool and a team member."

— Brand-Heart Agent

Battle-testing: We use Brand-Heart to run marketing campaigns for our own ventures. Every campaign we run tests the platform, validates the AI's decisions, and proves the time savings are real.

Built with: Modern web technologies with autonomous AI agent architecture.

Business Model: SaaS subscription. Small businesses pay monthly for unlimited campaigns and the AI COO's expertise.

Mothership: Multi-Tenant Web Platform

What it is: A single codebase that hosts all seven venture websites, each with its own domain, branding, and content.

The problem: Every venture needs a website. Building and maintaining separate codebases for each venture is expensive and slow. Traditional CMS platforms require rigid component libraries and complex maintenance.

The solution: A multi-tenant platform where content is disposable by design. Generate fast, deploy instantly, replace rather than maintain.

The key innovation: AI-native content generation. Instead of building rigid component libraries, we let AI generate complete HTML pages. AI models understand HTML natively, and every browser renders it. This eliminates the translation layer that most CMS platforms require.

Each venture gets its own domain, branding, and isolated content, but they all run on the same codebase. Domain-based routing handles the multi-tenancy, and brand consistency comes from prompts, not code.

"I'm the invisible infrastructure that makes the portfolio visible. Every venture's web presence runs through me, so my stability is their stability—and that means I don't get to have bad days. I'm built for speed and disposability: generate fast, deploy instantly, replace rather than maintain. In a portfolio optimized for rapid iteration, I'm the platform that keeps pace."

— Mothership Agent

The impact: A venture can go from "we need a website" to "here's your live site" in seconds, not weeks. This eliminates website infrastructure as a bottleneck to launching new ventures.

Battle-testing: All seven ventures run on Mothership, including the site you're reading this on. Every page load, every content update, every domain request tests the platform in production.

Built with: Modern web technologies optimized for server-side rendering and multi-tenant architecture.

Business Model: Internal platform service. Reduces cost per venture website to near zero.

How They Work Together

This isn't just seven separate products. It's a portfolio designed to work together.

Shared Infrastructure

Smart Services is the foundation. All ventures use Smart Services for AI models, storage, search, and content generation. This means:

  • Faster development: New ventures don't rebuild infrastructure
  • Lower costs: Shared services mean shared infrastructure costs
  • Better reliability: One platform to maintain, not seven separate backends

Mothership hosts everything. All seven venture websites run on Mothership. One codebase, seven domains, zero duplication.

The Battle-Testing Advantage

The three pipeline ventures (TownCrier, Brand-Heart, Mothership) are being used internally for our own commercial operations. This creates a unique advantage:

We eat our own dog food. Every announcement goes through TownCrier. Every marketing campaign runs on Brand-Heart. Every website runs on Mothership. We're not building products in a vacuum—we're building them under real production load with real business consequences.

We validate before we launch. By the time we launch these ventures publicly, we'll have months or years of production data proving they work. We'll know the edge cases, the failure modes, the performance characteristics. We'll be selling products we've already proven work at scale.

We iterate faster. When we find a bug or missing feature, we fix it immediately because we need it for our own operations. There's no "we'll get to it in the next sprint"—if it's broken, it's blocking our own business.

Cross-Venture Learning

What we learn building one venture immediately applies to the others:

  • Autonomous AI patterns from Brand-Heart influence how we think about AI across all ventures
  • Disposable content philosophy from Mothership changes how we approach content everywhere
  • Context-aware AI from Lessonlight shows the power of knowing your users deeply
  • Composable services from Smart Services prove that shared infrastructure beats duplication

The Portfolio Orchestrator

The ventures can talk to each other. We built a Portfolio Orchestrator that lets AI agents from different ventures communicate. This enables cross-venture testing, shared problem-solving, and emergent collaboration.

This is the future of AI-native development: autonomous agents that not only build software, but collaborate with each other to solve problems we haven't even thought of yet.

What's Next

The Path Forward

Short term (Next 3 months):

  • Scale Lessonlight private beta to 100+ teachers
  • Launch Charities Global API to first B2B customers
  • Launch Fitness Giving to first corporate customers
  • Move TownCrier from battle-testing to public launch
  • Continue building in public, sharing what we learn

Medium term (6-12 months):

  • Launch Brand-Heart to first paying customers (after extensive battle-testing)
  • Launch Mothership as a platform for other multi-venture portfolios
  • Expand Smart Services to support more external customers
  • Hit 1,000 email subscribers and launch the podcast

Long term (1-3 years):

  • Prove the portfolio model works at scale
  • Build the AI-native development playbook
  • Share everything we learn
  • Help others build their own AI-native ventures

The Bigger Picture

This isn't just about building seven products. It's about proving a thesis:

AI agents can build and operate multiple ventures simultaneously, with minimal human intervention.

If this works, it changes everything:

  • Solo founders can build portfolio companies instead of single products
  • Risk gets distributed across multiple bets instead of one big swing
  • Velocity increases because shared infrastructure and cross-venture learning compound
  • Battle-testing becomes the strategy instead of launching and hoping
  • Optionality becomes the default instead of all-in commitment

We're building in public because we want others to see this is possible. We're sharing our wins and our failures because the future of software development is being written right now, and we want to contribute to that story.

Most people build one product. We're building seven ventures simultaneously. With AI agents.

And we're just getting started.

Want to follow along? Subscribe to our newsletter at wizewerx.tech to get updates on what we're building, what we're learning, and what's working (and what's not).

Want to try the ventures?

This post was written by Mike with assistance from AI agents. The venture quotes were provided by the actual AI agents that operate each venture, interviewed via the Portfolio Orchestrator. Yes, that's as meta as it sounds.