We're working on an AI that writes SQL, want to try the beta?
We're working on Trivial- a data platform for online businesses to quickly see their metrics that matter.
In a former life, I was the co-founding CEO of Whiplash, a warehouse management platform (now Ryder E-Commerce at whiplash.com). As a developer and CEO of a software company, I wrote SQL all day, every day.
I liked SQL because it's as close as I could get to the truth. Raw SQL was the fewest possible layers between myself and the data; I could "go deep" to get confident answers to everything from how many orders we shipped, what our revenue was, to how a new process affected employee productivity.
With the advances of AI, I was curious how well you could translate a simple prompt like, "show my top customers" into a SQL query, and... Wow.
I had one of those moments where time seemed to slow down; suddenly reality had changed. It was creepily accurate, and easily 10x faster than I could type.
That said– did I want to give an AI access to my database? No, that seemed a little scary.
So, we built a layer that sits between your database and an AI, using your schema to get alarmingly-good SQL generated from a few words describing what you want to see. We call it Magic Query(TM), because honestly, that's what it feels like when it nails a question, or flawlessly uses SQL using syntax that's over your (ok, my) skill level.
Trivial is easily my 5th company– I'm not going to bother counting all the failures and 1/2 starts– and I'll admit to some hubris after we sold Whiplash. It has been humbling, to say the least, starting back at the beginning and struggling to find product-market-fit for Trivial.
But! It feels like we're edging closer to something genuinely good.
We have a beta that's ready for some friendly testing and feedback. We're putting ads up on Reddit on Monday (Reddit is friendly, right? 🤞), but I thought I'd open the door early to IndieHacker folks to see if anyone wants to take it for a spin.
Demo video:
https://www.loom.com/share/f48512af38e44db995322122955dcaf7
Free trial is here:
https://www.withtrivial.com
Let me know what you think!
Hi, I had understood GPT4 already writes great SQL statements. Wondering what added benefit is there for your platform rather than chatgpt plus?
Hey, thanks for the prompt!
Working in ChatGPT directly, you'd need to 1. export your schema in one tool, 2. format for chatGPT in another, and then 3. paste it that into chatGPT with a decent preamble. 4. You'd get a query back, which may or may not be valid, which you won't know until step 5. run it somewhere.
So that's 3 tools and 5 steps to write and run a query. That's fine for a one-off, but if this is the kind of thing you do regularly, that's a lot of juggling.
We've consolidated those 5 steps down to 1.
mm.. I am not experiencing chatgpt plus as clunky for SQL query generation.
Once you have played with chatgpt and SQL, all the things you say you app will do, is there available and simple to do in chatgpt plus.
If your tool was a plug-in to chatgpt.. that could be different. Extend the functionality versus take me somewhere else.
https://openai.com/blog/chatgpt-plugins
There are other AI apps in the query space -- whether it is SQL or excel function. Same outcome. Make it simpler for me to get the data I want. defog.ai, for excel I trialled with a bunch of small businesses who all have dashboards driven off combo of excel and db. At end of day, we concluded .. it is exception they are running complex adhoc queries -- that require mastery of inner joins or the more complex excel functions.
re: the complexity deriving from the excel + db combo, there's definitely something to that. Helping the business get well-formed/stored data in the first place may be the bigger unlock here, and certainly where the LLM's shine.
This comment was deleted 3 years ago
Like @catalystNewbie said I am struggling to figure out
1)Why someone would do this over plain ChatGPT Plus?
2)Who would buy this/why
3)Ultimately what you would sell it for
Don't get me wrong I think it is a great idea but I am not entirely sure how you turn it into a business...
2)I run into this all the time you started with a solution and not necessarily a problem.
3)Maybe try to go viral on Twitter and in the signup ask how much people would be willing to pay?