Infrastructure to connect LLMs and databases

Buster is an infrastructure platform that engineering teams use to connect their database to their LLM applications. We provide our customers with tooling for data cataloging, fine-tuning, security, evaluations, retrieval, sql generation, and more.

Team Size:3
Location:San Francisco
Group Partner:Jared Friedman

Active Founders

Blake Rouse

Blake Rouse is the co-founder and CEO of Buster. Previously, he led product development at a bootstrapped analytics startup (DataSpark, acqd by Threecolts). Blake is originally from South Carolina. He studied Product Management and Computer Science at BYU for three years before dropping out to start Buster.

Blake Rouse
Blake Rouse

Dallin Bentley

Dallin Bentley is the co-founder and CTO of Buster. Before starting Buster, Dallin worked as a Cloud Security Engineer. After that, he founded Leftovers (Surplus food marketplace) which was later sold. Dallin graduated from Brigham Young University with his Masters of Information Systems.

Dallin Bentley
Dallin Bentley

Company Launches

tl;dr: Buster gives developers everything they need to train LLMs to write accurate SQL queries for their database. Out of the box Buster handles fine-tuning, access controls, performance evaluations, deployments, and more. Watch a demo video here.

Hey, this is Blake and Dallin from Buster.

Buster is a developer tool for training Large Language Models (LLMs) to write production-ready SQL queries for your database. Buster provides the workspace for fine-tuning, model evaluations, managing access controls, and more.

  • Getting an LLM to understand the nuances of your schema is a nightmare
  • Pushing LLMs to production is a nightmare
  • Monitoring LLMs in production is a nightmare
  • Train an LLM to understand database nuances
    • Buster auto-generates documentation for your entire database
    • Buster utilizes a training pipeline to build out domain knowledge around your database
    • Buster fine-tunes an LLM, specific to your database, to recognize query patterns and understand nuances
  • Push LLMs to production
    • Buster hosts models for you
    • Buster includes a UI for the LLM that can be embedded anywhere as an iFrame (users can build dashboards/reports/etc using natural language)
  • Monitor and improve performance
    • Buster provides a monitoring space for engineers to track performance, usage, conversations, sql generation, user feedback, etc
    • Buster provides tooling to debug and quickly fix performance gaps
how it works
  1. Auto-generate documentation for your schema

  2. Train your model

  3. Embed “chat with your data” capabilities into your product


We’re live and have customers using Buster in production! You can request free access at buster.so

message for the skeptics

Many of you have been let down by text-to-sql solutions, that’s why we’re building Buster. If you’re a skeptic, we’re happy to connect you with current customers that use Buster in production :)

Other Company Launches

Buster - Enable your customers to chat with their data

Embed an LLM-powered reporting feature into your product, in less than an hour.
Read Launch ›