I have been following this board for about an year and this is my first post. I want to thank the contributors of this board and a special thanks to Saul.
Here’s what I think about AYX…
About myself: I’m a developer in the Seattle area and have worked for 14 years with one the greatest software company in the world ( you can guess it :)) developing some of the core software products/services with millions of users worldwide.
My expertise differs from Saul et al. in that I’m a hard core developer but very poor with business numbers ( i still don’t understand gaap vs non-gaap very well).
Having said that. Here’s why I like AYX.
A simple example without any technical jargon: A couple of years back i wanted to find out how many users out a total number were likely to return to use a service ( aka churn). To explain in the simplest terms here’s what I did.
-
Load the user data from multiple log files from many front-end servers to a no-sql DB.
-
Extract and schematize the unstructured data.
-
Persist that data in some DB ( can be relational.)
-
Run a pipeline to create historical data for some days (maybe months) so that we can have a range.
-
MOST Painstaking: Try to create some correlation using all the historical data from the above steps. This step brings in the infamous slow JOINs that no one likes (I’ll talk about MDB later someday ).
-
Load the data into a tool for some insights/analysis ( tools lile Tableau/Power BI)
[Most of the folks would be happy at this stage to glean important insights into their data.] -
BUT SOME STILL WANT MORE: How can all that historical data be of any use to us? How can we keep our customers? Which customer base looks the most attractive for our new feature? We need ML/AI?
-
This is where I end up working with Data Scientists( all PhDs). They use the data from Step 5. ( with a number of combinations) and apply their esoteric knowledge to build ML Models which can start providing better insights/predictions than Step 6.
My Take: There are a few ways you can achieve Step 8.
Work with Data Scientists like Step 8.
Use machine learning packages in R or some other library ( means employing someone like me with a hefty salary :)).
Use AYX ( at least from what I’ve understood so far) .
Although I haven’t tried using AYX, I know that they also help with some of the data prep stages ( think Stages 2 -5). So it’s not very surprising why AYX is doing well 
and why the CEO talks about citizen data-scientists.
Caveat: Pure SW companies who have many Data Scientists may not feel the need to use AYX. The question is do all businesses have Data Scientists + Data Engineers? It’s not difficult to figure out which option would be more expensive.
I’ve invested about 15% of my portfolio in AYX based upon the above experiences and not seeing any viable competition as of now. Please let me know if you think otherwise.
If you think this post was useful please let me know and I will be happy to share my thoughts about DDOG, since I’ve worked in that space too.
Cheers!
ron
developing some of the core software products/services with millions of users worldwide.