In 2018, I quit my job at Givaudan and joined Entrepreneur First (EF) with one goal: to build a startup.
For those who are not familiar with EF, they bring together a cohort of people with different backgrounds, some with deep technical expertise, some with domain expertise, and help them find co-founders and build companies.
You don’t need to come in with an idea. But as someone coming from the food and beverage industry, I felt that the main thing I could bring at that stage was my understanding of the industry and the problems within it.
So before the cohort started, I wrote down four or five startup ideas. All of them were in food, beverage or FMCG, and most had a deep-tech angle, partly because that was the lens through which EF would eventually decide which companies to back.
It is interesting to look back at those ideas now.
1. Rapid microbial testing for food
Traditional food testing in a lab can take two to three days. For a food company, contamination can have a huge impact, both on consumer health and on the brand.
The idea was to build a rapid testing kit that could detect microbial contamination in minutes rather than days.
This was actually the first idea I pursued when the EF cohort started.
I teamed up with a very smart biotech PhD who had worked on a technology that we thought could potentially be used to build this. We started speaking to customers, and the validation was very strong. This was clearly a real problem. In startup language, it was a proper “hair on fire” problem.
We spent around five weeks on it and became more and more convinced that the market need was real.
But somewhere along the way I started to realise that while it was a good startup idea, it might not be the right startup for me.
This was going to be a very research-intensive company. Commercialisation could take years. For the first few years, most of the hard work would be scientific research, and my role would probably be largely around fundraising and business development.
I like building things and seeing them move quickly. I wasn’t sure I wanted to spend the next few years waiting for the technology to become commercially ready.
So we decided to split up.
She later teamed up with another biotech PhD and continued working on the idea. They eventually raised funding, which also proved that the opportunity itself was very real.
It just wasn’t the company I wanted to spend the next several years building.
2. AI for manufacturing planning
Even in 2018, there were already quite a few startups applying AI to manufacturing.
But I felt there was an interesting gap in food and beverage manufacturing.
Food factories have their own complexities: allergens, sanitation requirements, multiple batch sizes, frequent product changeovers and a large number of variants. Production planning can get complicated very quickly.
My idea was to build an AI-based planning system specifically for food manufacturing.
I discussed it with a few people but eventually decided not to pursue it. There was already a lot of investment going into AI for manufacturing, and I felt it might be difficult to build something meaningfully differentiated.
Interestingly, I still think the problem is far from solved.
With agentic AI, there is now a possibility of moving production planning from something that happens weekly or monthly towards something much closer to real time.
So perhaps this wasn’t a bad idea. It was just an idea I chose not to compete on.
3. Cell-based meat
Just before joining EF, I was travelling in the US.
Beyond Meat and Impossible Foods had started getting a lot of attention, although their products were still available in relatively few places.
I found a restaurant in San Francisco serving an Impossible Burger and went there specifically to try it.
I was terribly disappointed.
The taste and texture didn’t work for me. But more than that, I wasn’t fully convinced by the proposition. If I wanted to eat vegetarian food, I was perfectly happy eating vegetables. Why would I want a highly processed plant product designed to look like meat?
Cell-based meat, however, made much more sense to me.
Instead of trying to imitate meat using plants, why not actually grow meat?
At the time, it was still a very new field, and as far as I knew nobody had built a serious cell-based meat company out of Asia.
My idea was to build a B2B cell-based meat company. That meant avoiding the additional complexity of building a consumer brand, and it was also much closer to my own background in B2B food sales.
The problem was that I needed a highly technical co-founder.
There wasn’t really anyone in the EF cohort with the right background for what I had in mind, so the idea never went very far.
Interestingly, Shiok Meats came out of the same EF cohort. Sandhya had to go and find her tech co-founder, Ka Yi, outside the cohort.
We all know what eventually happened to the alternative meat category. A lot of the initial excitement and investment disappeared.
But I still think cell-based meat will come back. The underlying logic hasn’t changed. The technology just needs more time to mature.
4. Predicting food trends using AI
This was the idea closest to my own experience.
I had spent more than seven years in the flavour industry and had seen first-hand how food companies decided what new flavours and products to launch.
A surprising amount of it came down to experience, intuition and gut feel.
At the same time, there was already an enormous amount of consumer data appearing online across social media, search, menus, reviews, recipes and many other sources.
My thought was quite simple: could AI use all this data to identify emerging food trends and predict where they were going?
The first version of the idea was narrowly focused on flavour trends.
But as we started speaking to customers, it evolved.
And eventually, that idea became AI Palette.
Looking back, the interesting part is that the first startup idea I worked on was probably a perfectly good idea.
We had customer validation. The problem was important. The technology was promising. The company later raised funding.
But I wasn’t particularly passionate about food safety, and I was learning the science of the problem as I went along.
So the second time around, I went in the opposite direction.
I chose the problem closest to what I already knew.
I had spent years sitting with food companies, helping them develop products. I understood how they thought, where the frustrations were and what kind of solution they might actually use.
That turned out to matter more than I realised at the time.
A startup idea can be good on paper and still be the wrong idea for you.
The idea I eventually built wasn’t necessarily the biggest market or the most technically impressive one on my list.
It was the one where I had a genuine advantage; and in the end, that probably mattered more.
Loading comments…