Dashboards say what happened. Daggy works out what to do next.
An optimization engine for what you spend and what you publish. Works with Claude, ChatGPT or your own agent.
Tested in public
AI agents forecast better with Daggy.
- With Daggy
- .228
- Prediction market
- .231
- Without Daggy
- .235
Four AI models forecast every NFL game before kickoff, each with and without Daggy. Each is scored by average Brier; lower is better. Over 121 games, all four did better with it.
About
Tristin Beckman founded Daggy after more than a decade of using data to answer messy real-world questions in both academia and industry. At Eventbrite he built revenue forecasts for the CFO, measured campaigns for marketing teams, and helped the company navigate its IPO and the COVID crisis. At Pendulum Intelligence he built, trained and deployed transformer-based models that uncovered risks to people, infrastructure and brand across millions of pieces of video and audio content daily. He has an MA in Economics and a PhD in Political Science with an emphasis in quantitative methods from UC Santa Barbara, and completed a postdoctoral research fellowship at Oxford.
From the blog
Synthetic agents are easier to manipulate in experiments than humans