← Back to Ideas

Wolfram Alpha for Databases

May 12, 2022

Wolfram Alpha is really incredible. It lets you type in just about any mathematical question, in plain english, and then it gives you an answer that is usually correct.

The Spark Notes Version

-----

This idea is a B2B SaaS product that would use the same concept as Wolfram Alpha.

It would integrate with your database (or whatever data backend you use). It would allow the non-technical teams in your firm (like marketing, product, sales, etc.) to perform database queries by writing in plain English what info they need.

Let's say a sales team wanted to know how much revenue was generated in the past week. Normally, they would first go to an analyst or a data scientist, who would write the query in SQL (or whatever language they use). Then, the analyst would send either the number or the full report. This is so many touch points for such a simple need. It also takes days from start to finish.

Imagine if the sales manager could instead just go into this program, type "sum of revenue in the past week" and get the answer.

Ideas like this, right in your inbox.

Let's be honest. You don't want to have to come to our site every time you want a new idea. Let us send ideas like this one right to your email, for free, every week.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

The Long Version

-----

Most database languages really aren't that far from English. If you don't know SQL, try to guess what the following will tell you:

SELECT email
FROM `customers`
WHERE total_spend > 20

Did you guess that it tells you the email of every customer who spent more than $20?

Probably.

Of course most queries aren't this simple, but they are usually at least similar to the plain english version.

This idea is to make a SaaS company that does what you just did, but in the opposite direction. It takes a human-readable input (like "how many customers did we onboard last week") and turns it into an SQL query, then returns the data. Basically an AI assistant to bridge the gap between technical and non-technical departments.

Okay, sounds cool. Is it feasible, though?

Probably.

We've seen recently that AI is shockingly good at interpreting what we need and turning it into code. OpenAI recently released Codex, which can correctly generate entire pieces of code from simple instructions almost 30% of the time!

Parsing a request into a structured query is a much simpler feat than that.

If you can get the tech side of this down, the marketing is almost done for you. Just hire some salespeople to bitch to B2B buyers. Purchasing cycle and compliance headaches notwithstanding, it is obvious how this program would benefit any company with separate marketing and engineering departments.

Engineering resources aren't wasted performing a simple query.

Marketing resources aren't bottlenecked by having to wait for the data.

All teams perform better -- they are more willing to take a data-backed approach, since they no longer have to wait forever to get the data they need.

This software increases revenue and decreases cost. There's hardly a better sales pitch imaginable.