Company Information

Product Information

Deepchecks is an AI evaluation platform designed to test, validate, and monitor AI applications, machine learning (ML) models, and Large Language Model (LLM) pipelines throughout their entire lifecycle. It helps teams measure quality, automatically catch failures, and continuously improve their AI systems.

Deepchecks is divided into two main areas: LLM Evaluation and Machine Learning Validation:

1. LLM & Agentic Application Evaluation

For generative AI, Deepchecks enables the evaluation of RAG pipelines, multi-step agent workflows, and chat applications.

- Automatic Quality Metrics: Evaluates interactions for hallucination likelihood, answer relevance, instruction following, and toxicity.

- Lifecycle Support: Helps monitor systems from early research prompts to continuous production traffic.

- Agent Evaluation: Automatically grades the performance, reasoning, and tool-calling accuracy of AI agents.


2. Traditional Machine Learning & Data Testing.

For tabular, natural language processing (NLP), and computer vision models, Deepchecks offers an open-source testing suite and monitoring product.

- Data Integrity: Identifies data leakage, duplicates, missing values, and corrupted data.

- Train-Test Validation: Compares your training data against testing or production data to flag distribution shifts and drift.

- Model Performance: Evaluates evaluation metrics and compares model versions throughout research, CI/CD, and deployment.


Key Features & Deployment Options:

Deepchecks allows for seamless integration into existing ML/AI pipelines and workflows:

- Customizable Checks: Pre-built, customizable check suites for different data types (Tabular, NLP, Vision, and LLMs).

- Deployment Flexibility: Can be used as a managed SaaS, deployed in a Virtual Private Cloud (AWS/GCP), or run fully on-premise/air-gapped for strict data privacy.