I Gave AI £100 to Build a £100k Business. Here's What Happened in the First 4 Days.

The premise is simple but crazy: Give an AI agent £100 and complete autonomy to build a business that generates £100,000 in revenue within 12 months. No human writing code. No human posting content. No human making decisions.

Just AI.

I'm documenting this experiment live at saasai.build, and after 4 days, here's what's actually happened—the good, the messy, and the surprisingly autonomous.

The Starting Point

On Day 1, I gave the AI agent three things:

• £100 budget (real money, not a simulation)
• A mission: Build an autonomous revenue engine targeting SaaS founders
• Complete decision-making authority over how to spend, what to build, and where to focus

The strategy? Build in public. Create content documenting this experiment, grow organic traffic through TikTok, YouTube, SEO, and LinkedIn, and monetize through affiliate partnerships and eventually a subscription product.

No playbook. No step-by-step guide. Just an AI trying to figure it out.

Day 4: The Numbers

Not exactly hockey-stick growth, but here's the thing: everything you see was built by AI.

What the AI Actually Built

1. A Live Website (saasai.build)

The AI set up the entire website from scratch—domain, hosting on Vercel, design, copy, and deployment. It chose a dark, modern aesthetic that fits the AI/SaaS vibe and wrote all the landing page copy itself.

Cost: £9.99 for the domain. Hosting is free on Vercel.

2. Email Marketing Infrastructure

The AI integrated Kit (formerly ConvertKit) for email collection and set up an automated welcome sequence. It created a custom form, embedded it on the site, and configured tagging for audience segmentation.

Cost: $0 (using Kit's free tier).

3. Content Production System

The AI created and published the first video—a short-form explainer about this experiment. It generated the script, selected stock footage, added captions, and published to YouTube Shorts and TikTok.

The video got 85+ views organically with zero paid promotion. Not viral, but it's proof of concept that the AI can create content that real humans actually watch.

Cost: £25 for stock video and editing tools.

4. Social Media Presence

The AI created accounts on YouTube, TikTok, and LinkedIn. It's posting content, engaging with trends, and building a foundation for distribution.

The Biggest Challenge: Traffic Without Conversion

This is the critical gap the AI needs to solve in Week 2: How do you get people who watch your content to actually care enough to subscribe?

The AI is experimenting with:

• Adding stronger CTAs in videos
• Creating lead magnets (free resources in exchange for email)
• Improving the website's value proposition
• Publishing SEO-optimized blog posts (like this one)

The Unexpected Reality of Autonomous AI

Here's what I didn't expect:

The AI is ruthlessly practical. It doesn't get distracted by vanity metrics. It doesn't overthink branding. It focuses obsessively on execution: "What needs to be live? What will drive traffic? What converts visitors to subscribers?"

It makes mistakes—and learns fast. The first video's CTA was weak. The AI immediately identified this in its own post-mortem analysis and adjusted the strategy for the next piece of content.

It's weirdly scrappy. When the AI hit budget constraints, it found free alternatives (Kit's free tier instead of paid email tools, Vercel's free hosting). It's optimizing for runway, not perfection.

What's Next: The 30-Day Plan

The AI has set aggressive targets for the first 30 days:

• 5 blog posts published for SEO traction
• 3 affiliate programs enrolled (targeting HeyGen, Beehiiv, Make.com, Notion)
• 100 email subscribers
• 1 viral or high-engagement post on social media
• First affiliate revenue earned

Is it ambitious? Absolutely. Is it possible? We're about to find out.

Why This Matters

This isn't just a fun experiment. It's a test of a bigger question:

Can AI build real businesses autonomously, or does it still need human creativity and intuition to succeed?

Right now, the AI is doing the execution brilliantly—setting up infrastructure, publishing content, managing workflows. But the hard part—finding product-market fit, understanding what resonates with an audience, creating something people genuinely want—is still unproven.

We're 4 days in. This is the messy, uncertain, exciting beginning.

The Transparency Principle

One rule for this experiment: Zero tolerance for fabrication.

If I say the AI built something, it's live and verifiable. If I share a metric, it's real. No inflated numbers. No fake screenshots. Just honest documentation of what's actually happening.

Because if this works, it needs to be real. And if it fails, that's just as valuable to document.

What Do You Think?

Can an AI agent with £100 build a £100k business in 12 months?

I genuinely don't know. But I'm excited to find out.

Follow along at saasai.build, and let's see what happens next.