A space for thinking, experimenting, and sharing on data and AI.
We publish ideas, frameworks, and small experiments that challenge assumptions, expose gaps, and make complex systems more understandable.
These are the domains we're actively exploring — where questions are more interesting than answers.
Moving from dashboards to decisions that matter. Understanding how data transforms into action.
Understanding when, why, and how AI adds real value to organizations and products.
Making large language models more reliable, transparent, and useful in real-world applications.
Turning questions into trustworthy, explainable queries that bridge human intent and data systems.
Measuring what matters and creating data that helps us test safely and effectively.
Monitoring model behavior, drift, and feedback loops to ensure systems remain trustworthy.
Ensuring the foundations of data remain strong amid automation and scale.
Building bridges between tables, text, and multimodal signals to unlock insight.
For over a decade, self-service analytics has promised data access for all. Text-to-SQL and natural language interfaces might finally deliver on that promise.
Read more →Moving beyond visualization to build data systems that directly enable and improve decision-making.
Read more →Understanding what it takes to measure trustworthiness, explainability, and practical utility in natural language data interfaces.
Read more →Derive from Data welcomes collaborations, questions, and small experimental ideas. If you're working on something that touches data, AI, or decision-making — or simply want to discuss an idea — reach out at hello@derivefromdata.com.