Review of the challenge - £1M in 100 days.
CJ Liu
Before London
I dropped out of Reed College in Portland, Oregon, in 2020, where I was an undergraduate majoring in computer science. I returned to Beijing, China, to start my own business.
Initially, from 2020 to 2023, I worked on a product called HiBug, which used AI to help programmers debug code, automate code reviews, and more.
Later, I pivoted to EdTech after a client at a college mentioned concerns about students using AI. We built the first AI model designed not to give students answers, called Shizhe AI (“Teacher AI” in Mandarin). My go-to-market strategy was to offer RMB 1,000 to users who could provide evidence that they had managed to get our AI to answer their questions. This helped us reach around 300,000 users.
In 2024, I introduced a new feature called “AI Student.” It put AI in the role of a student, asking children questions that they had to answer—for example, how to calculate the area of a triangle. The system rewarded users with badges and other incentives.
These innovative ideas attracted the attention of Xueda Group (www.xueda.com), one of the biggest education companies in China and a publicly listed company. They reached out to us.
We collaborated in several areas, including helping Xueda convert more users into customers, save teachers time, and improve the efficiency of grading papers.
Taking the Leap
In June 2026, I signed a contract to sell the AI product to Xueda and started a new challenge: going to London to make £1 million in cash in 100 days, starting with just £100. Alongside this, I wanted to become an influencer by posting videos and building an audience. I realised that personal influence was becoming more valuable with the rise of AI founders, so I needed to do something different.
The challenge was inspired by the Discovery reality show “Undercover Billionaire.” It’s a very famous show, but some viewers say it is scripted and that the production crew helps the participant find opportunities. My plan was to record everything, edit the footage, and post it online.
Everyone could see my live balance on www.day1cj.com. I connected the website to a bank account so that every penny I earned and spent would be publicly visible, along with notes I took during the challenge. I wanted to make the process as transparent as possible.
The challenge started on 24 June 2026. I realised it had already begun while I was on the plane. The most important first step was to connect with local people, so I struck up conversations with other passengers. These conversations gave me the impression that a significant proportion of people in the UK and Europe were not particularly fond of AI.
My idea was to print T-shirts with funny slogans, such as “Powered by coffee, replaced by ChatGPT” and “I hate my job, but even more when AI takes it over.”
The First Connections
I landed in London as a total stranger. I went to Imperial College London and spoke to young people while holding an iPad displaying the words “Looking for a cool co-founder.”
One person thought it was cool and introduced me to the “Pitch and Run” club, where founders and VCs run and talk every Thursday morning. I went along and made some really cool friends.
I shared my entrepreneurial experience at a company run by one of the club’s founders, helping high school students build projects. Another founder in the same building wanted to meet me, and we got along really well. He suddenly asked whether I wanted to meet his investor, Antler, whose office was a five-minute walk away.
I went over and spoke with a VP at Antler for an hour. That was on Day 3. I realised that connections and introductions were key.
The First Products
Later, I built a website called www.YinYangScience.com. The idea came from a YouTube video in which an Asian man offered traditional feng shui advice and fortune-telling on the street. The video had millions of views, which suggested that people were interested.
I thought I could turn this into a product, using AI to learn the rules of fortune-telling. I also thought people might feel a greater sense of privacy when talking to AI rather than another person.
I went to St Paul’s Cathedral with a sign saying “Free AI Fortune Telling,” and people queued up to try my product!
I edited the footage and posted the videos online. This brought traffic to the website, and I made around £1,000 in a week.
Housing was a huge problem, so I posted online asking whether anyone had a couch I could crash on for free. People were willing to help, mostly without charging me. They included international students who were away from London and people with a spare room.
Finding a Product People Wanted
I realised there were two big markets in London. The first was tourism: the city receives over 21 million visitors a year, mostly from the US and Europe, who stay for more than a week. June and July are particularly popular.
I built an app called DetectiveX, which lets users play a city treasure hunt. It gives clues such as “Find the Egyptian queen’s favourite cat in the British Museum.” Users can take a photo when they find it, and the algorithm checks whether they have found the correct object. They receive badges and credits for correct answers. I charged £14.99 for a week.
When I asked tourists for feedback on the street, I found that they were particularly interested in one feature I had added: showing the nearest toilets, places with air conditioning, Wi-Fi, and other amenities. Everyone I asked gave positive feedback on this feature.
I separated it into a new app called AhaThere. It costs £4.99 to download, with no further payment required.
My approach to promoting the app was also new: recreating videos that had already gone viral on Instagram, with AhaThere’s name included in the video. These are examples of the kind of videos I mean:
From Workflow to Client
Later, I built a workflow that used AI to find potentially viral videos online, which saved me time.
I shared this idea with another entrepreneur at Pitch and Run. He was a marketing expert and was really interested in it. After several meetings, he asked me to help his company build an AI marketing workflow.
It was highly customised work, but he paid me £5,000 upfront to help cover my living costs while I worked on it. This was around Day 60.
I spent the final month working on-site and developed a deep understanding of the team’s daily routines.
The client did not want me to post anything about the project online. However, he was willing to pay £35,000 in total if I could deliver a practical, productive AI + SaaS tool suited to his team of around 40 people and teach them how to use it.
On Day 97, he paid the remaining £30,000.
Where the Challenge Landed
That brought the challenge’s total revenue to £37,685.99.
The transaction history is available on www.day1cj.com.
Because of the NDA, I can’t share information about this client. This was also my first experience working as a forward-deployed engineer (FDE).
Here are 10 things I learned from these 100 days:
- Don’t be afraid of anything.
- Trust is expensive, and it will become even more expensive in the AI era. Build trust today.
- Software will change the world. Software is changing the world. But the value of software is going to vanish soon.
- Be clear about your focus: AI combined with an industry. Know which industry you are focusing on.
- Be ruthless in pursuing your goal, because it is possible.
- You may doubt yourself, hesitate, or feel depressed for a while. But, overall, work through it and stay positive.
- Your feelings matter.
- Hard work is necessary for success. Making the right, smart decisions is harder—and much more important than working hard.
- Attention is expensive. Value your reputation.
- Make connections and be an extrovert, even if you don’t enjoy it. You can’t deny that extroverts get to enjoy what the world offers sooner and communicate more effectively. If you don’t like it, treat it as a job you have to do.