TechPrufer
Get started
Mistral AI
Mistral AI logoMistral AI · Feature trail

Search Toolkit

3 posts · Apr 2025 – Aug 2026

What this trail covers

Mistral AI introduces Search Toolkit, a composable framework for building production search pipelines for AI applications. It integrates ingestion, retrieval (BM25 sparse, dense embedding-based, and hybrid), and evaluation (recall, precision, MRR, NDCG) into a single framework with a shared interface. The toolkit supports custom document formats, preprocessing, and advanced retrieval features like LLM query rewriting and reranking. It is battle-tested across various verticals and integrates with LLM-as-a-judge evaluation frameworks, including structured outputs for RAG evaluation.

All Mistral AI trails
  1. Introducing Agentic Search | MistralThis post introduces Agentic Search, an enhancement to the Mistral Search Toolkit. It details a multi-step retrieval loop that allows AI models to interact with data using five specific tools: `search`, `open`, `navigate`, `read`, and `grep`. The post provides concrete benchmarks on FinanceBench and OfficeQA Pro, showing improvements in correctness (e.g., 26.7% to 86% on FinanceBench) and reductions in latency (p90 drop from 255s to 154s with navigation) and token usage (e.g., -23.9% for MM 3.5 with navigation). It contrasts this with traditional one-shot RAG, highlighting its limitations with long, dense, or multi-document queries. The core technical contribution is the introduction of these agentic tools for more sophisticated document navigation and information retrieval.
  2. Introducing Search Toolkit | Mistral AIThis post introduces the Search Toolkit, a new framework for building production search pipelines for AI applications. It integrates ingestion, retrieval (BM25, dense, hybrid), and evaluation (recall, precision, MRR, NDCG) into a unified system. Key technical contributions include a common configuration interface across modules, support for custom document formats and preprocessing via adapters, and built-in evaluation metrics to measure retriever performance independently. The post also highlights its use in enterprise search, RAG, and domain-specific retrieval, and its integration with Connectors for agentic workflows.
  3. Evaluating RAG with LLM as a Judge | Mistral AIThis post details the application of Mistral's structured outputs feature to enhance LLM-as-a-judge evaluations for RAG systems. It outlines how structured outputs can enforce clear schemas and criteria for evaluating context relevance, groundedness, and answer relevance, making LLM-based evaluation more reliable and machine-readable. The post provides a Pydantic schema example for RAG evaluation and links to a GitHub notebook demonstrating the implementation.