Launching today
DecisionBox
Autonomous AI Discovery For Your Data — Open Source
30 followers
Autonomous AI Discovery For Your Data — Open Source
30 followers
DecisionBox is an open-source project and connects to your data warehouse, runs autonomous AI agents that write and execute SQL, and surfaces validated insights and actionable recommendations, without you asking a single question.
🔍 The Problem
We've been working with countless customers on data analytics for many years, and every data team hits the same wall: extracting insights from data is manual, slow, and completely single-threaded. PMs wait days. Analysts write one-off queries. AI chatbots only answer the questions you think to ask. Nobody is proactively looking at your data and telling you what you're missing.
🤖 How It Works
DecisionBox fixes that. Connect your data warehouse — BigQuery, Redshift, Snowflake, DataBricks, PostgreSQL and more — and AI agents autonomously explore your data. No prompting, no question-writing. A single discovery run executes 100+ SQL queries, forming hypotheses, cross-validating findings, and producing structured insights ranked by severity with affected user counts, confidence scores, and specific recommended actions.
✅ Built for Trust
Every finding goes through two independent validation passes against your warehouse. Every SQL query and reasoning step the agent takes is logged and visible. The output is not a chat response — it's a prioritized list of validated findings your team can act on.
🧩 Industry-Aware Intelligence
We built DecisionBox as an open source project with a domain pack architecture. Domain packs give agents industry-specific strategy — what to look for, what matters, how to analyze it. We're launching with E-commerce, Gaming (Idle, Casual, Match-3) and Social Network (Content Sharing) domain packs, with more coming soon. The community can contribute packs for SaaS, fintech, or any vertical — or you can create your own for your specific use cases.
🔒 Your Data, Your Infrastructure
DecisionBox agents run inside your infrastructure — queries execute against your warehouse, in your cloud account. No data extraction, no third-party storage, no vendor lock-in. We ship Terraform modules for GCP and AWS, and Helm charts for Kubernetes, so you can go from zero to a running cluster with a single terraform apply, or start with Docker Compose locally.
👋 Who We Are
Our founding team has been building tools and solutions for data and analytics for over a decade, including four years at AWS (Amazon Redshift). We come from a startup culture — we were co-founders and founding team members at DataRow, which was acquired by Amazon in 2020.
🚀 Get Involved
We'd love your feedback. Star and fork the repo, run a discovery on your own data, or build a domain pack for your industry. PRs are welcome — and let us know what you find.