Matih
Plain-English questions, Trusted insights from your data
Start new thread

Matih - Plain-English questions, Trusted insights from your data

by•
Text-to-SQL tools are schema-blind. they guess the joins and hand you a number you can't check. Matih builds a context graph of your data:-the entities, how tables relate, what each metric means, then reasons over that ontology to answer. Ask in plain English. Matih writes the SQL and auto-routes it to the right engine- DuckDB for a spreadsheet, Spark or Trino for billions of rows- from an Excel file to a full warehouse. You get the answer next to the SQL. Trusted insights, no black box.

Add a comment

Replies

Best
Maker
📌
Hey Product Hunt 👋 Every "chat with your data" tool I tried had two blind spots. It could read my column names but had no clue what my business actually was:- ask "which customers are at risk?" and it'd confidently join the wrong tables. And the moment the data got big, it fell over. A good analyst is useful for the opposite reasons: they carry a model of the business in their head — what an "account" is, that refunds live in a different table than orders, what "active" really means here. and they don't care whether the answer lives in a 200-row sheet or a billion-row warehouse. So we built both into Matih. It maps your data into a context graph: the entities, the real relationships, the metric definitions; and grounds every answer in that ontology instead of guessing from column names. And it scales without you thinking about it. Ask a question and Matih writes the SQL, then auto-routes it to the right engine - DuckDB for a quick spreadsheet, Spark or Trino when it's billions of rows. Same question, same plain English, whether it's a CSV you dragged in this morning or your entire warehouse. Then it show its work: the answer next to the SQL, the rows behind it, the reasoning. Wrong? You see exactly where and fix the graph - you don't re-prompt into the void. It's live at . I'll be here all day - tell me the parts that annoy you, that's the feedback I actually want.
Maker

Excited to finally share Matih with the Product Hunt community! 🚀

We started building Matih because every data team we spoke to had the same problem: business users wanted instant answers, while data teams were buried under repetitive ad-hoc requests. Existing AI tools could generate SQL, but not trust.

We'd love your feedback—whether it's on the product, the UX, or the idea itself. What would convince you to trust AI with your business data?

sounds very interesting, ai into academic! congrats on the launch!

Maker

 Thanks so much! Really appreciate the support! 🙌

Maker

 Thank you!!

Finally tried this on our messy Snowflake schema and the context graph piece is genuinely useful, not just buzz. Loved that it shows the SQL right next to the answer so I can actually verify the join logic instead of just trusting a number.

Maker
Thank you! That means a lot. Our goal has always been to make AI answers transparent and trustworthy, not something you have to take on faith. Really appreciate you trying Matih and sharing your feedback! 🙌