Nums AI
Nums AI provides Causilo, a foundation model API for structured data that predicts classification and regression outcomes directly from user-provided context tables without a training step. Users send a table with known answers plus new query rows, and the API returns predictions, eliminating the need for feature engineering or model infrastructure. The platform offers separate models for classification and regression, with cell-based pricing that counts context and query rows together.
- Artificial Intelligence
- Developer Tools
- Software Only
Funding
Raised to date
₩6BRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
Est. $4.5Macross 2 rounds
BVKISV₩2B · Pre-Seed Tranche
Announced Nov 2026 · from article date
Disclosed cumulative: $4.5M
Founders
Founder details are not available yet.
Product
Problem
Traditional machine learning on tabular data requires significant feature engineering, model selection, and training infrastructure. Teams often struggle to build accurate predictive models quickly, especially when dealing with small to medium-sized datasets or when they need to iterate rapidly on new prediction tasks.
Solution
Nums AI offers Causilo, a foundation model API for structured data that predicts outcomes directly from user-provided context tables without requiring a training step. Users send a context table containing known answers in the target column along with new query rows, and the API returns predictions for classification or regression tasks. The platform hosts the model, eliminating the need for users to manage infrastructure, and provides a simple Python client for integration. Causilo supports both classification (causilo-clf) and regression (causilo-reg) models, with the ability to infer class labels, probabilities, means, medians, and quantiles.
Target Audience
Data scientists, machine learning engineers, and software developers who need fast, accurate predictions on tabular data without the overhead of building and training custom models.
Features
- Zero-training-step prediction: directly infer target values from context tables without model training
- Dual model support: causilo-clf for classification and causilo-reg for regression tasks
- Python client library (causilo-client) for simple API integration
- Usage tracking with GET /v1/usage and request ceilings via GET /v1/limits
- Cell-based pricing that counts context and query rows together, excluding the target column
- Stored contexts to avoid re-uploading the same data for repeated queries