On May 23rd, 2018 Lifecycle, together with another 6 companies, graduated from the 1st cohort in Bucharest chapter of Founder Institute. It was a fun journey!
I’ve got into Founder Institute by almost an accident. They hosted the first meetup I came since I moved to Romania. I was always passionate about entrepreneurship. So it didn’t take me long to decide to join the program. It was a perfect occasion to start my company in Romania and to get to know the local entrepreneurial community.
What business idea to choose?
When I was starting in Founder Institute I had 2 projects in mind:
– automation of target audience (TA) research
– my old project InfoSpace – a generalized web-development framework for arbitrary information objects – which I wanted to revive.
The story of the first one is quite simple. While I was a stay-at-home mom, I’ve been working as a freelance internet marketer. Once I’ve got stuck with the development of target audience for car detailing business, in which I have no expertise. I put a lot of effort, collected and organized all type of information about potential customers. But I couldn’t get good enough understanding what drives people to buy car detailing in 2 weeks. Sure enough, I’ve lost the client.
But I discovered that I really enjoyed the process of collecting information and getting insights about target audiences. I’ve seen a lot of opportunity in focusing exclusively on target audience research. It requires a lot of manual work and time, and as a consequence, it is not done thoroughly by many of my internet marketing colleagues.
The second project InfoSpace was on pause for 8 years since I’ve got my first child. If I would know what I learned in Founder Institute back then, probably I would have continued development. But at that point, I was using graphs as the base structure and got stuck with limited possibilities of MySQL… And just about a month before starting Founder Institute program I discovered graph databases and Neo4j. I was so excited because now I had a professional and easy to use tool for working with graph objects.
The next step was to implement knowledge graphs. Years ago I tried to study Natural Language Processing. But I couldn’t – it didn’t make much sense. The time I went to Founder Institute meetup, I was reading a book about math. structure of natural languages. And it did make sense for me. I was quite excited to try to model text as a knowledge graph, to develop Natural Language Understanding module.
Doing Natural Language Processing without expertise in the field, sounded very complicated. So I decided to develop automated target audience exploration as a business idea. That’s how my company Lifecycle started.
Ups and downs
The Lifecycle idea was very simple originally. I wanted to collect general information about professions:
– what people do on their work
– when they start and finish the working day
– how much they earn
– where they learn
And about private life:
– activities with kids
And then, using statistical clues, to combine different pieces together, reconstructing a model of human life. Not a real person’s life, but a realistic avatar’s life.
At Product Development Assignment I realized that it is not simple at all. It requires a lot of data and Natural Language Understanding (NLU) module to collect and organize it. I’ve got scared. The NLU problem looked quite complicated and almost impossible to master. But my business advisor convinced me not to drop out and not to pivot. NLU thus become an integrated part of Lifecycle.
The good part about Founder Institute is that they don’t let you relax. Assignments are very tough in the beginning. And before the second Mentor Review, there was almost no chance to actually work on the implementation. I had to develop the business idea, without going in-depth of the technical part.
After the second Mentor Review, even though most of the mentors were very supportive of our business idea, I felt depressed. The next 3 days were the scariest. My business idea sounded like a science fiction or a dream and I had no idea how to code it.
Once I started development, I quickly realized that I have a general structure of the algorithm. It sounded like I’ve proven a mathematical theorem. My enthusiasm got back. And by graduation the structure of the algorithm developed further, uncovering amazing insights about natural languages. With every day it looks more and more solid. So it is only a matter of time when it is going to be coded and to what extent of perfection we can get. Would our algorithm be able to understand exactly what people say? That’s what we aim for.
Living our dreams with Founder Institute
Without Founder Institute I wouldn’t be able to develop NLU algorithm that fast, if at all. And definitely not in such a short time. The directors and mentors, tough assignments, group support – they all make such a creative environment, that our ideas progress really fast. I live my dream now!