What did I learn after launching a product without a laborious customer discovery?

#buildinpublic #founderjourney #founderstories

A few months ago, after burnout, a machine learning bachelor's degree diploma, and many changes in my life, I started building Docstract's MVP with Bianca Necula. We were super excited and put a lot of time into tech development because it was our comfort zone.

Docstract.com is a generalist product that helps you extract data from bulk unstructured documents and export it as CSV. (for example, you have a lot of receipts and you need to extract key information and then export it as CSV)

The differentiator is that you can define your own document model from our UI without coding skills (you can say I want this field and this field) and without training. (demo here: https://www.youtube.com/watch?v=HY-EcVABHMk)

But let's talk about what I learned these months.

Docstract's story began a few months ago when one of our clients from my web development agency (inntech.dev) wanted to streamline the HR onboarding process, and we created an interface where employees upload their documents, we extract data from images with proper technology and then insert that data through an API. Then we realized that we could do that from unstructured documents as receipts through LLM & OCR. We built the receipt parser prototype and realized that we could create an interface to define your custom model and extract bulk.

The technology part was perfect. We had a nice accuracy, a nice UI, a good response time and so on. But we fell in love with the product and didn't realize that we needed an ICP, a specific market, and so on, because without this stuff, it was impossible to scale.

After finishing the MVP, we went to a lot of conferences and events like HowToWeb, Saastock Dublin, and so on and we started talking with investors, consultants, and potential customers (even with people who wanted to buy Docstract) we found that we needed to step back and make some changes.

Even if my bachelor's diploma was about "How you can build an MVP in a technology start-up," where I wrote about pre-sale methods, validation methods, and so on, the practice was different. The programming part drove us; technically, everything was excellent, but we forgot some things, such as:

To think about what type of product we want to build. Is it a huge one backed by VCs or a lifestyle product?
To make the customer discovery process in a correct way without biases;
To start with a niche, get some clients from this niche, and then expand;

We stepped back and analyzed all this stuff and now have a discovery strategy; we stopped our tech development and started building a genuine tech product for a specific market.

Now, we're looking at how we can implement our technology to streamline expense management (especially document storage, classification, easy search, and easy sharing to accountants) in SMBs (small & medium businesses).

Thanks to all my supportive people who helped me understand that I can bring more value in business contexts without specifically writing code. That helped me to get my head out of my laptop and think of solutions to build and scale tech products.

I'm building my career and products in public, so you can follow me on Twitter (https://twitter.com/iandreileo) and LinkedIn (https://www.linkedin.com/in/ilie-andrei-leonard/)