End-User Business Intelligence (BI) and Analytics is a major thesis and investment focus area for WestWave Capital. When I first joined WestWave Capital, it was an area that I got increasingly excited about and started actively searching for potential investments that would meet our thesis.
So it is with sheer delight that I am happy to announce that our first investment around this , MachEye, came out of stealth recently. MachEye is led by a brilliant visionary, Ramesh Panuganty, and more on their recent launch can be found here.
My own experience as a business executive who was thirsty for analytics esp. around my sales and marketing data taught me the hard way that Business Intelligence and Analytics, while always driven by the business, has been the domain of data analysts and technical engineers. Business users have to increasingly rely on technical teams to be able to make sense of their data. These projects can take weeks if not months using Legacy BI products.
In my particular case, as my boss (the General Manager of our business) got increasingly frustrated at the lack of insights that could be directly obtained from our CRM and marketing tools, we had to implement Tableau to create analytical dashboards that could be shared with the executive staff. The exercise to pull data from different sources (ETL — extract, transform and load), then to setup the right queries (in SQL) that mapped what our GM wanted and to generate the right dashboards took approximately 3 months and required consulting with Tableau experts.
In the age of Google and Alexa, interfaces where we can basically ask what we want for using natural language search and query, BI really shouldn’t be this hard. There are two major problems with legacy BI that have prevented us from getting to this panacea. These are the areas where we think innovation can make BI much more friendlier for the business user.
- Front-end: Most BI tools require a technical team to translate the actual business query (often expressed in English or your preferred natural language by a business user) to a SQL query. Things often get lost in translation and the queries have to be continually corrected, managed and updated. This is not an agile cycle resulting in long times to gratification and a very static experience.
- Back-end: Data may not reside all in the same place and even before you can query it, it has to be ETL-ed into a central database. Data is often not clean and complete. This has to be fixed and the data then needs to be centrally reconciled so that SQL queries can cut across multiple data-sources (by doing a cross-join in this centralized database). This can take weeks to setup and may require serious database schema wrangling.
These are the two major impediments for end-user BI. MachEye solves exactly these two problems. Note that it is necessary to solve both of these problems but solving each on its own is not sufficient to really make BI end-user ready. MachEye solves the above two problems as follows:
- Front-end: In MachEye, end-users don’t write SQL, they write natural language. For example, “what were my top performing car sales across all dealerships in California”. With legacy BI, this would typically be translated to a SQL statement by a technical team supporting the business end-user. Arguably, this would be a simple SQL query. However, very soon, quite predictably, the end-user would probably want additional insights such as the top performing car models, which dealerships had the highest sales, highest growth and were there any outliers, etc. etc. — for example did a particular car model stand out at a particular dealership. Now it starts getting much harder in SQL. MachEye however guesses all of this and provides these additional insights along with the results of the original query. And as if this wasn’t enough, to round off the natural language experience, the results are also presented in natural language as a short 60 second audio-video presentation that you can email to your boss! Impressive!
- Back-end: MachEye doesn’t require you to duplicate data or go through lengthy ETL. You get to directly query existing on-premises or cloud databases. The cloud-based SaaS solution is extremely easy to get started with. If you already use a cloud database such as SnowFlake, you can be setup and running your first query within a few minutes as opposed to a setup of weeks required with Legacy BI.
No wonder, when we first saw the MachEye solution, we were ultra impressed. The big picture and big vision matters a lot to investors but little things go a long way in acting as strong signals for investors too as to why a particular startup is the right one to back. Several such things stood out for us when we dug further deeper into MachEye. Just some examples of such things were:
- The MachEye solution is not a complete black-box unlike other ML/AI solutions for analytics. The user can actually see the SQL statements that are generated and even modify them if needed. Seems like a trivial point, but an important one. How does one even know that the result of a natural language query is correct? What if the answers that you are getting were wrong all along the way? With MachEye you can always fall back and check the SQL to make sure that it captures the intent of your original query.
- Most natural language queries rely on just some tags and do not catch the language nuances ending up generating the same answer to different queries. Example: What were the sales last month vs what are the sales by month should result in very different answers. Surprising that many solutions give the same answer to these two different queries.
Be sure to check out the MachEye website.
One would have thought that the BI space is done and dusted and there is no more innovation to be done! But once you see MachEye, I guarantee that you will come to believe otherwise.