# Lightning Rod Labs > Predict Anything with AI This file provides information about Lightning Rod Labs to help large language models understand and reference this publication's content. ## Posts - [AI Reaches Human SuperForecaster Range](https://blog.lightningrod.ai/p/ai-reaches-human-superforecaster-range): On ForecastBench, the best AI systems are now statistically tied with the human superforecaster median — and one of them is our compact Foresight-v3. - [Foresight V4: Better Predictions, Cheaper Inference](https://blog.lightningrod.ai/p/foresight-v4-better-predictions-cheaper-inference-ccc7): Foresight-V4 sets the Pareto frontier for AI forecasting - [Predicting Patient Outcomes From Clinical Notes](https://blog.lightningrod.ai/p/predicting-patient-outcomes): A small AI model, trained on existing hospital notes with no manual labeling, predicts patient outcomes like ventilation, dialysis, and mortality — and runs on a single GPU. - [Forecasting Geopolitical Risk with Foresight Learning](https://blog.lightningrod.ai/p/forecasting-geopolitical-risk-with-foresight-learning): A small open model, trained from public news, massively outperforms GPT-5 at predicting military strikes. - [Turning Fed Beige Book PDFs Into A Calibrated AI Economic Forecaster](https://blog.lightningrod.ai/p/turning-fed-beige-book-pdfs-into-a-calibrated-ai-economic-forecaster): Predicting changes in regional economic conditions using only timestamped PDFs - [From Metric to Prediction: Forecasting Supply Chain Disruptions with Foresight Learning](https://blog.lightningrod.ai/p/from-metric-to-prediction-forecasting-supply-chain-disruptions-with-foresight-learning): Predicting high-impact events from noisy unstructured data - [How We Built the #1 AI Forecaster](https://blog.lightningrod.ai/p/how-we-built-the-number-1-ai-forecaster): Foresight V3 is #1 on Prophet Arena, beating every frontier model. Here's how a small team with a small model outperformed the largest AI models in the world. - [Build a Labeled Forecasting Dataset from Real-World News in 10 Minutes](https://blog.lightningrod.ai/p/build-a-labeled-forecasting-dataset-from-real-world-news-in-10-minutes): Use Lightning Rod to turn real news into a model-ready prediction dataset. - [AI-Driven SEC Risk Prediction: Separating Signal from Noise](https://blog.lightningrod.ai/p/ai-driven-sec-risk-prediction-separating-signal-from-noise): Transform 10-K narratives into calibrated probabilities using Foresight Learning. Outperform general purpose models like GPT-5 with specialized risk prediction. - [Using the Future to Train Prediction Models](https://blog.lightningrod.ai/p/using-the-future-to-train-prediction-models): Real world event resolution as a scalable source of supervision - [Foresight-32B Beats Frontier LLMs on Live Polymarket Predictions](https://blog.lightningrod.ai/p/foresight-32b-beats-frontier-llms-on-live-polymarket-predictions): A forward-looking test of AI forecasting performance on live prediction markets ## Pages - [](https://blog.lightningrod.ai/) - [archive](https://blog.lightningrod.ai/archive) ## Authors - [Zack Johnson](https://blog.lightningrod.ai/authors/zack-johnson) - [Ben Turtel](https://blog.lightningrod.ai/authors/ben-turtel) ## Categories - [AI Research](https://blog.lightningrod.ai/t/AI Research) - [Prediction Models](https://blog.lightningrod.ai/t/Prediction Models) - [Tutorial](https://blog.lightningrod.ai/t/Tutorial) - [Training Data](https://blog.lightningrod.ai/t/Training Data) - [Forecasting](https://blog.lightningrod.ai/t/Forecasting)