When I was at FreeCharge, I did a lot of data engineering. One of the recurring patterns that I saw was:
1. A new product or feature gets launched.
2. The PM wants to schedule a report for her feature.
3. The report should run at a particular frequency.
4. The PM along with others from the C-team should get this report in their e-mail inbox.
We had hundreds of reports like this running at FreeCharge. We explored tools like Redash, Metabase etc; but all these tools were geared towards creating beautiful dashboards and visualizations, report generation and delivery through email was given a step motherly treatment. We hacked together a solution to solve this problem with bash/Python scripts, GIT, Jenkins etc. The query to generate the report was written by the business analyst team but they had to depend on the engineering team to schedule the report, convert the report into HTML and deliver it over email. One of the goals of the solution was to make the business team self-reliant in report generation. Since the solution was stitched together using disparate tools that were not really meant to solve this problem, it sort of worked but had its own chinks. This seeded the thoughts for Kwery in my mind.
Kwery is a tool that solves the above niche problem really well. It was also an attempt at building a single person lifestyle business that could supplement my regular income. I followed a lot of lean startup principles while building Kwery. Built an MVP, deployed it at an organization that had the above pressing problem and then iterated from there.
With Kwery, initially, I had to take a call as to whether to make it a SAS product or an in-house deployment. I decided against the SAS route as I was not sure how many organizations would be comfortable giving complete access to their data sources to an external fledgling SAS tool. I wanted to make the onboarding process of Kwery as simple as possible. Hence, I shunned all external dependencies for Kwery, opted for an embedded database so that I could package Kwery as a single binary. The only dependency needed to run Kwery is Java 8 runtime.
I did deploy Kwery in a couple of places which use it even today. Kwery has received a lot of love from these places but I never hit the numbers that I had in mind when I started. The opportunity cost was bearing on me. I had to take a call between continuing working on Kwery or cut my losses and move on, I chose the latter.
Like the old saying – Every cloud has a silver lining, I have open sourced Kwery under MIT license. It is very easy to get started with Kwery. Give it a spin, open Github issues if you face any trouble. Pull requests are very much welcome.