From day one we knew we wanted to create a no-code platform more powerful than a Jupyter Notebooks and much faster and intuitive than Tableau or PowerBI. What a challenge! We wanted to make something really disruptive not another incremental boring BI or Notebook or boxes interconnected!

Our main references were Trifacta (for no-code/low-code data wrangling), CARTO (as no-code data visual analytics for maps), Ayasdi (for advanced visual data analytics for explainable machine learning models), Quid (for no-code visual data analytics) and Datawrapper and Flourish for interactive data visualisations. We also had many great ideas from academic papers about HCI for data analytics Juan knew.

Each of these companies had raised 100M so we were not that naives, we to knew we couldn’t build all of that with the 500K we raised from Kfund.

So we started by focusing on an MVP with the most differentiate UI, that would define the DNA of our vision. We started with the Graph for clustering text and unstructured data. It was also the hardest interface to build, that would help us to think about the kind of architecture we needed to build this kind of tool.

At the end it took us almost 2 years to have something that would be able to work with a few thousand rows datasets, but we were able to start selling it as something very cool (nice to have) for marketing agencies reports clustering tweets.

So right after that we focused on other features for interpreting the clusters like the cross fitters, the Compare section. Then we re-architecture the whole thing with Apache Arrow to be able to support our own data types and they hybrid architecture that would allow us to scale the concept from few thousand rows, just last year to a few millions now and very soon to hundred of millions and define any complex data pipeline with a simple low code DSL that would also allow us to support templates for the Wizard.

We also started the Plot section at the beginning of this year, which is key to be able to visualise very basics charts. Like a box plot. We are still missing many but we will have them before end of year. Also the ability to customize these charts even more than Datawrapper or Flourish, which is key for communication/presentation.