Excited to share a demo video created by one of our users in the baseball industry, Jonathan Moye! This demo uses real data that baseball organizations analyze everyday. Check it out here: https://lnkd.in/eRf4qujf Some soundbites: "Baseball organizations today work with extremely large databases that in the past have required someone with advanced SQL skills to return meaningful data. The problem is, not everyone working in baseball has these skills. DaLMatian gives those without SQL skills the power to query data." "The coolest part about DaLMatian is that it connects directly to your organization's Slack, Microsoft Teams, whatever communication platform you use." "A user who is interested in finding a list of players with 20 home runs and 20 stolen bases in a season could ask this question directly in Slack chat, just like they were communicating with a data analyst, and this list would be returned right into Slack."
About us
DaLMatian empowers data teams to focus on what matters most by answering ad-hoc questions from business stakeholders instantly.
- Website
-
https://www.dalmatian.ai/
External link for DaLMatian
- Industry
- Technology, Information and Internet
- Company size
- 2-10 employees
- Type
- Privately Held
Employees at DaLMatian
Updates
-
We launched on Product Hunt! https://lnkd.in/eTgbk-9W
-
Check out our Y Combinator launch!
DaLMatian (YC W24) empowers data teams to focus on what matters most by answering ad-hoc questions from business stakeholders instantly. Co-founders Alan Du, Jason C., and Derrick Xin are pooling together their experience in analytics at Riskified, AI at Google Research, and engineering at Google & Duolingo to solve a burning pain point for companies: expanding data accessibility and analyst bandwidth. With DaLMatian, business stakeholders get their questions answered immediately, while data analysts cut out time spent on ad-hoc questions and instead focus on the highest impact analysis that directly drives revenue growth. Getting started in their on-prem/local application takes <5 minutes. For any database, any size, any level of complexity – simply open a file of past queries to get started. The Data Analyst Language Model (DaLM) then learns the business logic behind the queries and improves with more use. Congrats to the team on the launch – and welcome to YC!
Launch YC: ⚡️ DaLMatian - Supercharge your data team | Y Combinator
ycombinator.com
-
At DaLMatian, data security is a top priority. That's why we designed an application that is on-prem/local - we do not track anything. Furthermore, no actual data from your database leaves your environment. Schema and query code are sent to the LLM, but no actual data. Any PII in query code is masked as well. If you are interested in trying our product but concerned about security, we are happy to share more about why we fall into the "low risk" category from a security standpoint.
-
-
Excited to announce that DaLMatian now supports data warehouses of any size! We have designed a fully scalable solution tested on tens of thousands of tables. We support all major databases: Snowflake, SQL Server, BigQuery, Postgres, MySQL, SQLite, and more. To get started, you simply need to connect your database and open a file of past queries in our IDE. We do all the heavy lifting from there - from storing your schema to parsing each query and extracting the business logic. The key to our fully scalable solution is a retrieval augmented generation (RAG) system that supports retrieval of both structured and unstructured sources of context. Structured sources include your schema and question-query pairs gathered from usage in the IDE. Unstructured sources include unlabeled queries parsed from messy scratch files. Why is RAG the key? Because for enterprise-scale data teams, we're talking schemas with hundreds of thousands of columns and scratch files with thousands of queries, which cannot all be fit into an LLM's context window from both a context limit and cost effectiveness perspective. Our RAG system finds the needles in the haystack and reduces the inputs for a given question down to an amount that is both far below the context limit and efficient in token consumption. And it's lightning fast to ensure a great user experience: <0.3 seconds in our enterprise-scale testing environment!
-
Check out our origin story!
Wanted to share some of my motivations behind starting DaLMatian. 2.5 years ago I took a leap of faith and joined Riskified as a data analyst. I call it a leap of faith because prior to this I was deeply entrenched in the world of quant trading and didn't really know that much about a data analyst's role in a tech company and the value they bring. I learned very quickly that data analysts at Riskified are critical to the bottom line - we quite literally were measured on the bottom line, since our primary KPI was the profitability of our ML models. A major challenge I faced that inspired DaLMatian was that I was spending large chunks of my day writing SQL queries to answer ad-hoc questions from business stakeholders, taking away time that could've been spent on directly improving the bottom line. In the world of ecommerce fraud detection, we described this as spending too much time summarizing the fraud as opposed to stopping it. Turns out this kind of problem is not at all unique to Riskified. The need for increasing the bandwidth of analytics teams, and giving business decision makers more insights faster, is shared across many industries. Many even face this challenge to a greater degree than we did. I hope this sheds some light on why we're working on DaLMatian! If this resonates with you or anyone in your network, I would love to connect.
-
Check out our demo video to see DaLM (Data Analyst Language Model) in action! Sign up for access here: https://lnkd.in/eMUqbdHi
DaLMatian Demo
https://www.loom.com