Issue #94: 19 AI Business Ideas, Tactics, Strategies Worth Stealing

Good morning.

I've been talking about the one-person business for a few years now.

The idea that a single operator, armed with the right AI tools and workflows, could compete with agencies. That headcount would become a liability, not an asset. That the economics of business were about to flip upside down.

For a long time, I felt like I was shouting into the void.

Not anymore.

The past month, I noticed something. Founders, investors, analysts, and operators are all converging on the same conclusions.

So I did something I don't usually do. I stopped writing about how to build and started collecting what others are seeing.

This issue is different. It's a curated intelligence briefing. Nineteen powerful ideas circulating right now that directly relate to scaling your business with AI and agents.

Some of these will validate what you already know. Some will challenge your assumptions. A few might change how you think about the next twelve months.

Here's what the smartest people are saying about AI business right now and what it means for you.

— Sam

IN TODAY’S ISSUE 🤖 

• The Solo Agency Revolution

• AI as Personal Infrastructure

• Documentation as Competitive Advantage

• Speed vs. Craft

• Post-Singularity Moats

• Real-World AI ROI

• The Clout Economy

• The AI Opportunity Gap

• Human Data as a Trillion-Dollar Market

• AI Boom vs. Dot-Com Boom

• Everything is Computer

• AI-Powered Content Transformation

• The "Real Marketing" Counter-Revolution

• The AI PM: From Translator to Intent-Former

• AI-Powered Mentorship

• The Human Element as Differentiator

• Data-Powered AI Agents

• Synthetic Customer Research

• The Agentic Commerce Infrastructure

Let’s get into it.

1. The Solo AI Business 

The prediction landed this week: a single person "living in the terminal" could run an agency generating $100,000 per month with just four clients.

That's a 25x productivity multiplier compared to traditional agency structures.

Read that again. Twenty-five times.

The traditional agency model required bodies. More clients meant more employees. Growth meant overhead. Profit margins compressed as you scaled.

AI agents break that equation.

One technically proficient operator can now deliver what used to require a team of fifteen. The bottleneck isn't human resources anymore. It's workflow automation expertise and the strategic deployment of agents.

The competitive advantage has shifted. Team size used to signal capability. Now it signals inefficiency. The agencies that will dominate aren't hiring. They're building agent-driven workflows that multiply individual output by orders of magnitude.

If you're running a service business, audit which functions can be replaced by agent-driven workflows. The savings aren't 10-20%. They're 80-90%—while maintaining or improving output quality.

2. AI as Personal Infrastructure

Something interesting is happening with tools like Claude Code. They're evolving beyond their original purpose into comprehensive personal assistant platforms.

Users are creating sophisticated ~/agents folders with skills and commands that sync across multiple AI platforms. They're building custom infrastructure for their AI-powered workflows.

This is the emergence of a new category: personal AI infrastructure.

Think about what this means. Instead of using AI tools as they come, operators are building standardized agent configurations that can be deployed across use cases. Like code libraries, but for AI capabilities.

The businesses that invest in building this kind of modular, reusable infrastructure will gain compounding advantages. Every new workflow builds on the last. Every agent configuration can be adapted and deployed in new contexts.

The fast-changing landscape requires continuous adaptation. But the operators who treat their AI setup as infrastructure (not just tools) will pull ahead.

3. Documentation as Competitive Advantage

Claire Vo documented 106 individual AI workflows with step-by-step instructions and visual infographics.

The effort required? "A lot of tokens, manual editing, and a dash of patience."

This is labor-intensive work. And that's exactly why it's valuable.

Most businesses treat AI workflows as tribal knowledge. Someone figures out how to use Claude for customer research, and that knowledge lives in their head. They leave, the knowledge leaves with them.

The businesses that win will treat AI workflows as intellectual property. Documented with the same rigor as software code. Version-controlled. Continuously improved. Transferable.

This transforms tribal knowledge into assets. New team members can onboard in days instead of months. Scaling becomes a matter of deploying documented workflows rather than hoping institutional knowledge survives.

Establish a workflow library. Document prompts, screenshots, exact configurations. What feels like busywork now becomes competitive advantage later.

4. Speed vs. Craft

John Maeda raised an important question this week: while AI accelerates the think-make-think loop, does faster iteration deepen craft? Or do we need to consciously preserve slower, contemplative cycles?

This tension matters for anyone implementing AI in their business.

Speed is not always the goal.

AI can dramatically accelerate execution. Research that took days takes minutes. First drafts appear in seconds. Iteration cycles compress from weeks to hours.

But strategic thinking doesn't always benefit from acceleration. Brand development requires contemplation. Creative work needs incubation. Some decisions get worse when you make them faster.

The businesses that thrive will develop frameworks for this. "Acceleration zones" where AI should maximize speed. "Contemplation zones" where human judgment and deliberate pacing create value.

Not all work should be optimized for velocity. Knowing the difference is a skill that matters more as AI gets faster.

5. Post-Singularity Moats

In an AI economy, what creates defensible competitive advantage?

The framework is elegant: moats must trace back to atoms and real-world assets.

Three sources of defensibility exist:

World Data: Unique datasets. Customer data, proprietary research, specialized information no one else has. Strong moat.

Process Data: Methodology and domain expertise. How you do what you do. Weak moat—AI can eventually replicate process.

The Work Itself: Compute resources and execution capability. Commoditized. No moat.

Generic AI cannot replace domain experts because it lacks process data. But process data alone is vulnerable. The strongest position combines all three: proprietary data, specialized methodology, and execution capacity.

Audit your competitive position across these dimensions. If you only have process knowledge or execution capability, you're vulnerable to commoditization. The defensible position is owning unique data while using AI to amplify everything else.

6. Real-World AI ROI

Here's a concrete case study that should get your attention.

A Slack-based human-in-the-loop AI agent achieved a 33x return on investment in its first year. Millions in ROI from a $60,000 investment. Six weeks to develop. Minimal ongoing costs: $1K/year in AI expenses, minimal infrastructure.

The key success factor wasn't full autonomy. It was human-in-the-loop design.

The system handled complex decisions while maintaining quality and oversight. Users worked with it in Slack—where they already spent their time. No new interface to learn. No behavior change required.

This is perhaps the most powerful validation of the bionic business thesis: properly implemented AI agents deliver extraordinary ROI with modest investment. But "properly implemented" means human oversight at critical decision points, especially in early deployments.

Prioritize human-in-the-loop agent designs over fully autonomous systems. You'll get faster time-to-value, reduced risk, and organizational trust in AI systems. The Slack integration pattern is particularly effective—meet users where they already work.

7. The Clout Economy

The rise of TBPN (Technology Business Programming Network) demonstrates something important about the current moment. The line between influencer and entrepreneur is blurring.

The hosts' approach—treating content creation with the same intensity as building a startup—has made their podcast a status marker for Silicon Valley's elite. Guests include Mark Zuckerberg, Marc Andreessen, and Sam Altman.

Attention and access are becoming as valuable as technical capability.

This isn't about becoming an influencer. It's about recognizing that content and community are growth channels, not marketing afterthoughts.

The businesses that document their AI implementation journey, share workflows, and build in public create network effects. They attract customers, partners, and talent through visibility. They build moats through community that competitors can't easily replicate.

Consider content and community as core growth strategies. Not something you'll get to later. Part of how you scale.

8. The AI Opportunity Gap

Survey data from Lenny Rachitsky reveals massive demand gaps between current AI usage and desired applications across professional roles.

The biggest opportunities aren't in output tasks. Users have figured out AI for writing and basic coding.

The gaps are upstream:

• User research for PMs: +28 percentage point demand gap

• Prototyping across all roles: +24-27pp

• Post-code work for engineers (documentation, code review): +24-25pp

• Strategic thinking for founders (product ideation): +29pp

This is a roadmap.

Users are hungry to apply AI to messy, strategic, upstream work. Research. Prototyping. Strategic analysis. Quality assurance. The businesses that solve for these high-demand, low-supply use cases will capture significant value.

Stop optimizing already-solved output tasks. Focus on moving AI upstream in the value chain—where the demand gaps are largest and the competition is thinnest.

9. Human Data as a Trillion-Dollar Market

Here's a counterintuitive thesis: as AI automation scales, human expertise becomes more valuable, not less.

The prediction? Structured human judgment and expertise will become a $1 trillion annual market.

Why? AI systems require continuous human grounding. Demonstrations. Evaluations. Corrections. Every AI model needs human data to train, validate, and improve.

This reframes the relationship between humans and AI in business.

Human labor becomes a revenue-generating asset. Every hour of work can simultaneously run operations, train AI models, and generate additional revenue. Every customer service interaction, every design decision, every strategic choice—captured and structured properly—becomes training data.

Design workflows that capture human decision-making in structured formats. This transforms operational work into dual-purpose activity: delivering immediate value while building long-term AI training assets that become competitive moats.

10. AI Boom vs. Dot-Com Boom

The current AI boom is fundamentally different from the dot-com era.

The dot-com boom ran on debt. The AI boom runs on balance sheets.

Major technology companies are financing this transformation with strong cash positions. Not venture debt. Not speculative investment. Real money from profitable businesses.

This matters for long-term planning.

The AI revolution isn't a fleeting trend. It's a structural change in the economy with sustainable financing. The infrastructure is financially stable.

Make long-term investments in AI and agent-based automation with confidence. This isn't going away. The businesses that treat AI as a long-term infrastructure investment—not a short-term experiment—will compound advantages over those still testing the waters.

11. Everything is Computer

A16z published an analysis this week arguing that modern technology has converged on the smartphone as a fundamental paradigm.

Electric vehicles, drones, robotics—they're all variations of the same electro-industrial stack. Different form factors, same underlying architecture.

The insight matters because it highlights what's actually scarce.

Software and AI capabilities are abundant. What's scarce is the "modular middle"—the suppliers that can build and integrate the components of this electro-industrial stack.

For businesses scaling with AI, this is a reminder that software alone isn't the whole picture. The hardware and manufacturing ecosystems that underpin AI matter. Understanding the entire stack, from silicon to software, creates advantages that pure software plays can't match.

12. AI Content Transformation

A practical example of AI's power landed this week: the Acquired FM podcast transformed into a 300-page physical book using Claude AI.

The narrative arc is preserved. The key insights retained. The format was completely transformed.

This demonstrates something important about content economics.

AI enables transformation across mediums. Podcast to book. Book to course. Course to workshop. Blog posts to documentary scripts. The same intellectual property can now exist in multiple formats with dramatically reduced production costs.

This democratizes publishing. Individuals and small businesses can create high-quality content products that were previously only possible for large publishing houses.

Leverage AI to repurpose existing content into new formats and products. Every piece of content you create is now a seed for multiple revenue streams and audience expansion opportunities.

13. The "Real Marketing" Counter-Revolution

As the world drowns in AI-generated content, a powerful counter-trend is emerging.

A return to the real.

The prediction for 2026: standout marketing will come from doing real things. Putting in real effort to create durable things with taste and craftsmanship. Showing real evidence for outcomes. Designing real-world events. Forming real relationships.

The winning formula combines these real-world connections and human craft with a "terminally online" presence. Both, not either-or.

This is a critical message. AI should amplify human authenticity, not replace it.

The businesses that will break through the noise are using AI to automate the mundane. Freeing up human talent to focus on high-value, high-craft activities that build actual customer loyalty.

Your AI handles the mechanical work. Your humans do the work that matters.

14. The AI PM: From Translator to Intent-Former

The role of the Product Manager is undergoing a fundamental shift.

Previously, a PM's primary job was translation. Take customer needs. Translate them into engineering requirements. Manage the gap between what users want and what developers build.

That translation layer is compressing.

As AI agents become capable of taking a well-formed problem and producing working code, the PM's job changes. The spec is becoming the product. The time between "I know what we should build" and "here it is" has collapsed.

This makes the strategic work of knowing what to build more important than ever.

The bottleneck is no longer implementation. It's clarity of vision. The ability to define problems and articulate intent with the precision required for an AI agent to execute.

This isn't just about PMs. Every business needs this skill now: forming intent clearly enough that an agent can act on it directly. Vague thinking becomes visible immediately when agents can't execute on it.

15. AI-Powered Mentorship

Imagine having Charlie Munger as a strategic advisor.

Not the real Munger. A digital mentor trained on his vast writings, interviews, and mental models. Capable of reading millions of times more information than a human. Providing investment or business advice from his unique perspective.

This is happening now.

AI agents trained on the accumulated wisdom of legendary business and marketing leaders. Democratizing access to world-class mentorship. Any entrepreneur can tap into mental models that previously required proximity, luck, or millions in consulting fees.

The frontier extends beyond public figures.

Companies are building AI mentors trained on their own internal experts and historical data. A scalable way to preserve and distribute institutional knowledge. When your best salesperson retires, their wisdom doesn't walk out the door.

For operators, this opens new service opportunities. Building and deploying AI mentors that capture expertise most companies let evaporate.

16. The Human Element as Differentiator

As AI automates the mechanical aspects of work, the most durable competitive advantages will be rooted in deeply human elements.

A few principles worth internalizing:

"Loving attention beats LGTM culture." The work that wins is made with care, not shipped with indifference.

Leaders should be "fork-shaped", possessing deep expertise in multiple areas, not just one spike.

And perhaps most important: "You can't compete with somebody having fun."

These principles provide crucial counterbalance to the pure automation narrative.

AI is a tool, not the end goal.

The most successful bionic businesses will use AI to free up their teams for the things AI can't do. Developing taste. Fostering a joyful and excellent culture. Applying genuine, loving attention to craft.

The best work feels like play. No amount of automation can replicate the power of a team that is genuinely having fun.

17. Data-Powered AI Agents

AI agents are moving beyond simple task automation into strategic intelligence work.

Manus AI's partnership with SimilarWeb demonstrates what this looks like in practice. AI agents can now access 12 months of web traffic history, benchmark competitors instantly, analyze marketing channels and traffic sources, and get regional traffic breakdowns. All powered by trusted digital intelligence.

This is data-powered decision-making at scale.

Businesses can now automate competitive analysis, market research, and strategic planning tasks that previously required expensive consulting or manual research. The work that used to take weeks of analyst time happens in minutes.

This validates the thesis that the biggest opportunities lie not in automating output tasks but in automating intelligence gathering and analysis. AI agents are moving upstream into high-value strategic work.

18. Synthetic Customer Research

Here's a development that should change how you think about market research.

A new paper demonstrates that LLMs can predict real purchase intent with 90% accuracy by impersonating customers with demographic profiles, evaluating products, and having another AI rate the responses.

The methodology is remarkably simple. Ask an LLM to impersonate a customer. Give it a product. Collect impressions. Have another AI rate them. No fine-tuning or training required. Beats classic machine learning methods.

This represents a paradigm shift in product development.

You can now simulate customer reactions, test product concepts, optimize pricing, and predict market demand without the time and expense of traditional market research. "Test" with synthetic customers before investing in real customer acquisition.

The risk and cost of product launches just dropped by an order of magnitude.

19. The Agentic Commerce Infrastructure

The final piece may be the most transformative.

Google's Universal Commerce Protocol (UCP) is an open-source standard designed to power the next generation of "agentic commerce." It enables AI agents to seamlessly discover products, check out, and complete purchases on behalf of users across any platform.

The collaboration is massive: Shopify, Etsy, Wayfair, Target, Walmart, and over 20 global partners including Adyen, American Express, Mastercard, Stripe, and Visa.

UCP solves the N-to-N integration problem by creating a single integration point for all consumer surfaces. The protocol standardizes the full commerce journey—from discovery and consideration to purchase and order management.

This is a watershed moment.

It creates a "write once, sell everywhere" ecosystem where products and services can be discovered and purchased by any AI agent—Google, ChatGPT, Claude, whatever comes next—without building custom integrations for each platform.

Businesses that adopt UCP early will gain first-mover advantage in conversational commerce. This channel is poised to be as disruptive as mobile commerce was in the 2010s.

Nineteen ideas. Three threads.

The first: the economics of business are inverting. Solo operators achieving agency-level output. 25x productivity multipliers. The bottleneck shifting from implementation to clarity of vision. Human expertise becoming more valuable as AI scales, not less.

The second: as AI automates the mechanical, the human becomes the differentiator. Real marketing beats AI-generated slop. Loving attention beats LGTM culture. You can't compete with somebody having fun.

The third is about infrastructure. AI agents are moving upstream into strategic intelligence work. Synthetic customers can predict purchase intent with 90% accuracy. Universal commerce protocols are creating "write once, sell everywhere" ecosystems. The plumbing for an AI-native economy is being laid right now.

These threads aren't in tension. They're complementary.

I've been saying for years that AI would enable one-person businesses to compete with large organizations. That headcount would become a liability. That the operators who mastered AI workflows would pull ahead of those who just hired more people.

That's happening. The proof is everywhere now.

But the operators that are winning are using automation to free themselves for the work that matters. The craft. The real relationships. The joyful attention that no agent can replicate.

And they're paying attention to the infrastructure being built. The protocols, the data integrations, the commerce standards. Because the businesses that plug into this infrastructure early will have advantages that compound.

The future belongs to a Bionic Businesses, like yours. Human judgment and AI capability are integrated so seamlessly you can't tell where one ends and the other begins. Achieving outcomes neither could accomplish alone.

That future is here.

Until next time,
Sam Woods
The Editor

.