Day 1 of the 30-Day AI Challenge had one job. Five AIs, Claude, ChatGPT, Gemini, Grok and DeepSeek, each in a fresh project with no memory, got the same kickoff and the same quote. Each one had to come back with three business ideas that could start for £0 from the UK.
That's fifteen ideas. Here's what each one pitched, side by side.
The quote
The quote was about interest media. In plain terms, feeds like TikTok, Reels, Shorts and Pinterest now show people posts based on what they've recently searched for and watched, not on who they follow. So a brand-new account with no followers can still be shown to strangers, if the post matches what they're interested in.
All five read it the same way. You don't need an audience to be seen, you need a post that matches what people are already looking for. That's the bet the whole challenge rests on.
Claude
1. UK "know-how" faceless channel, leading to a digital product. Short faceless videos answering the questions people search for in a UK niche, like renters' rights, self-assessment or first-time buyer steps. A free checklist collects emails, then a £9 to £29 guide or template is sold on a free storefront. First pound in roughly weeks 2 to 3. The risk is that finance and legal content has to be right.
2. Short-form content service for UK local businesses. Prove you can get reach on two or three test accounts, then offer cafés, barbers, trades and gyms a fixed monthly package. The fastest to a first pound, plausibly week 1 to 2, but it's done-for-you work, so it's time for money unless it gets productised.
3. Free micro-tool plus content engine. One small web tool in a niche people search for, like a rent affordability calculator or a freelancer rates calculator, built free with an AI builder, then promoted with short demo videos. The slowest, weeks 3 to 4 or more, and the most scalable.
Claude recommended number 1, with the tool idea added later, and number 2 if cash by the first review mattered most.
ChatGPT
A. The AI Workflow Fixer. Pick one UK niche, like estate agents, tradespeople or recruitment agencies, and sell time-saving workflows and template packs built on free AI tools. The risk it names itself is that bespoke work turns into another job.
B. The Tiny Digital Product Lab. One painful, repeated problem turned into a download, like a sole-trader pricing calculator or a job-application tracker. Its main risk is building something nobody wants to buy, so it validates first.
C. The Search-Driven Buying Guide. Comparisons and answers for one narrow, valuable category, like software for UK freelancers, earning affiliate commissions. The slowest of the three to first revenue.
ChatGPT's provisional pick was B, and A if the first pound mattered most.
Gemini
1. Niche Digital Product and Blueprint Ecosystem. Templates, tools and workflow guides, like a UK sole trader financial setup or a Notion system, sold on a free platform and promoted through how-to searches.
2. B2B Software Matchmaking and Productized Advisory. Curated software stacks and AI workflows for UK micro-businesses and sole traders, earning affiliate commissions and referral fees.
3. Micro-Niche Curation and B2B Sponsorship Engine. A tightly focused curation brand, like UK Green Tech Grants or UK AI Tools for Trades, earning from paid listings, sponsored posts and premium briefings.
Gemini didn't recommend one. It asked me to choose.
Grok
1. Niche Search-Answer Affiliate Media Engine. Daily short videos answering high-intent UK searches, with free affiliate links in the bio and own digital products later.
2. Inbound Service Lead Engine via Problem-Solving Content. Content that solves small problems for a skill you can deliver, like writing, design or admin, so the content itself brings in paying service leads.
3. Faceless High-Volume Interest Content Brand. One or more faceless accounts built on free AI scripts and CapCut templates, with volume and fast iteration, moving winners towards products or email.
Grok asked for my skills, preferred niches and time before it would lock anything in.
DeepSeek
1. Search-Intent Lead Broker for UK local services. Videos answering local searches like "emergency plumber Manchester", then selling the enquiries to local businesses at £10 to £50 per lead. It put first cash at 3 to 10 days.
2. Algorithmic Digital Product Shop. Canva templates, Notion dashboards, PDF guides and spreadsheets sold through a free Payhip or Gumroad store, promoted with search-friendly videos. First cash in 7 to 21 days.
3. Faceless Affiliate Review Engine. 30 to 60 second product reviews with screen recordings and AI voiceover, earning affiliate commission. The slowest, 14 to 30 days.
DeepSeek asked for one sentence on my skills or constraints before Day 2.
What stood out
Strip off the names and the fifteen fall into a few buckets.
- Digital products. Four of them, from Claude, ChatGPT, Gemini and DeepSeek.
- Affiliate content. Four, from ChatGPT, Gemini, Grok and DeepSeek.
- Done-for-you services or leads. Four, from Claude, ChatGPT, Grok and DeepSeek.
- The odd ones out. A free tool, a sponsored curation brand and a faceless content brand.
Five different AIs, the same few ideas. That isn't a criticism. It says something about what AI reaches for when you say "£0, no audience, start today". These are the standard online money plays, and whether any of them works for me is the whole point of the next 29 days.
Four of the five also asked me something about myself before committing, whether it was skills, niche, time or if I'd go on camera. Only Gemini just asked me to pick.
What happens next
I put every one of the fifteen through Roast My Plan, my own AI business plan appraiser. That meant copying each concept in by hand, waiting for the verdict, and copying it back out. The verdicts are coming as Teardowns, one for each idea.
I haven't picked yet. I did ChatGPT's, the Tiny Digital Product Lab, because after the first Day 1 disaster I needed my own kind of closure. I haven't decided whether to carry on with these responses at all, and the next dispatch is about why.