How to find ideas

Over the last few months, many people have asked me what it’s like to leave a full-time job to start something. People are curious about what it’s like to go head first into something without knowing exactly what you’re going to build.

I’ve decided on the area I want to build in and launched a product: Kili. It feels like a good time to share a little bit about my journey and how I got here.

I hope this helps anyone else trying to look for an idea. Whilst most of this is related to building software, I think it can apply to many other things: building a different kind of business, writing etc.

The spark

I left Deliveroo in April 2022 with the desire to build something. In reality, I’d been dreaming about it for years. I’d had the itch before joining Deliveroo, but it’s really Deliveroo that fuelled the fire.

In my previous job, I’d use a software called Stata to write economic reports and run regressions. I joined Deliveroo and realised no one had even heard of it. I started learning SQL because I wanted to have data at my fingertips.

One thing led to another and I found myself write code in Javascript for an hour or two every day. At the time, it felt like a futile exercise on most days: what was the point of this? was I really going to build this myself?


Anyway, it was fun and I really enjoyed being able to build something from scratch. I kept going, but rarely showed the output to anyone other than my wife. Imposter syndrome for sure. I’d change this if I could, and put more stuff out into the world. It’s uncomfortable as hell but you have to do it. It never goes away, but it does getter with practice. The same thing happened with this newsletter by the way. The first essay was tough to get out. This essay is still hard, especially because it’s personal, but not as hard as the first one.

Eventually, I started putting stuff out there. There’s an extensive list of projects that I built, shipped and killed. A non-exhaustive list:

  • Invoice management for freelancers
  • Workout tracking
  • Product management tooling
  • Crypto-first link in bio

It’s rare to land on something great on your first try. We apply this to everything else: learning a language, a musical instrument or cooking, it really is no different in software.

Building and failing is natural. Treat it as a part of the journey and get stuff out there.


Once you find something, you will get excited about it for a couple of weeks. Once that fades, you will start to question it. I’ve found this pattern with pretty much every idea I’ve come up with.

Persistence alone can make all the different. The invoice management software was never going to be the next Salesforce, but if I had stuck with it, it might have been running on auto pilot and generating a few thousand dollars a month.

You need some balance here. Persisting through ideas that will never work is a futile effort. But there’s no magic answer here and in some ways, that is the beauty of building something. Two ways to de-risk this: talk to your potential customers every week, and if you can, get them to start paying.

Persistence is can make or break your product.

Kind of business

There are many kinds of businesses you can build. You can build a 1 person business, a lifestyle business or a venture scale business. Outside of the first one, it can be hard to choose between a lifestyle and a venture scale business.

If you prefer to have full control over your business and are happy to dedicate many years to growing it slowly, a lifestyle business might be the right choice. Most businesses are intended to be lifestyle businesses. You fund your business using some combination of money from friends / family and revenue from customers.

If you believe you have unique insight and want to build a very large business, a VC backed business might be for you. The question you will get asked is: “how big can this really get?”. This is because of venture capital works. VC returns follow a power law distribution, meaning that a small number of bets return very large returns.


To make their business model work, venture capitalists look for opportunities that can become very large and give you lots of capital to go work on it. As a rough benchmark, in SaaS, you should ask yourself whether you can hit >$100m ARR.

Mission or obsession

To build something truly meaningful, you need to be able to obsess.

It will be easier for you to identify an area if you have a mission you are passionate about, or you have an area you can obsess about. An example of the former is addressing climate change. An area might be something like cryptocurrency or artificial intelligence.

When I left Deliveroo, there were two areas I was very bullish about: crypto and AI. I spent a lot of time in the former and nearly started something there. I quickly realised that there was lethargy. As a person, I need to see my product in the hand of users or businesses. It was difficult to achieve that in crypto unless you were building specific products (collectibles, NFTs, decentralised finance etc).

Once I started researching and building in AI, I felt a pull. I found my self obsessing about the tech and what it could achieve. I’d look forward tor reading papers about the latest research in AI. I’d write code to understand how something worked or what was possible.

Anyhow, I say this because there are days when you feel like you are hitting a brick wall. A mission you believe in or an area you are excited about will help you get through those days.

Launching Kili

The first AI product I built allowed you to query databases with natural language. The pain point was allowing anyone, technical or otherwise, to get the data they need. After thinking about it for while, I felt like AI could be much more powerful.

The ability for AI to retrieve and summarise information is incredible. I built a prototype using my own newsletter content and it was fantastic. In building it, I realised there were lots of thorny problems: importing your data sources, managing them, analytics for Q&A etc etc. And so, Kili was born.

Kili’s mission is to 10x the productivity of every employee using the power of AI. We’re starting by building tools for sales, customer success and customer support. Our early customers are using the product to save time and money on support. Their customers are able to get responses faster, and their agents are able to increase the number of queries they can respond to.

It’s early days but I truly believe we are only scratching the surface of what’s possible with AI. Today, AI is summariser and content generator. I’m excited about a world where AI is an action taker. It allows us to do more meaningful work and I’m excited about that future.

If your company wants to explore reducing your support costs whilst delivering an exceptionally customer experience, please get in touch.