If you're building agents that work with tabular data (sales pipelines, customer data, inventory, financial records) you've probably hit this: agents spend tokens generating ML code that doesn't work, or produce unreliable results.

TabPFN MCP gives LLMs 2 tools: fit_and_predict (fits a model and runs predictions) and predict (uses a previously fitted model). Your agents don't need to impute missing data, encode categorical features, or preprocess messy tables as TabPFN handles it natively.

Available on ChatGPT, Claude, n8n and other major LLMs. Uses streamable HTTP for broad compatibility.

Keeping the beta small to work closely with early users. If you're shipping with structured data, join here: https://priorlabs.ai/deployment/model-context-protocol

Disclosure: I work at Prior Labs.