Berserk is a unified telemetry platform that ingests logs, metrics, traces, and AI model output without requiring predefined schemas.
Funding
Funding not disclosed
Founders
Product
Problem
Organizations deploying AI models generate large volumes of text-heavy logs alongside traditional metrics, traces, and telemetry, often requiring separate systems to store and analyze each data type. This fragmentation hampers correlation, increases operational complexity, and drives up storage costs, especially at petabyte scale.
Solution
Berserk offers a single telemetry platform that ingests logs, metrics, traces, and AI-generated output without requiring predefined schemas. Built in Rust, the engine stores data on object storage and provides fast, KQL-based querying across all data types. By consolidating observability data, Berserk enables users to correlate events, performance metrics, and AI inference details in real time. The platform is optimized for high‑volume, text‑heavy logs while remaining cost‑effective even when handling petabytes of data.
Target Audience
Primary customers are DevOps, SRE, and AI engineering teams that need a consolidated observability solution for large‑scale, AI‑driven applications.
Features
- Schemaless ingestion pipeline supporting logs, metrics, traces, and AI model output
- Unified query language (KQL) for cross‑type correlation and analysis
- High‑performance Rust engine designed for low latency at petabyte scale
- Object‑storage backend that delivers affordable, durable storage for large text datasets
- Real‑time indexing and retrieval optimized for text‑heavy log streams
- Built‑in support for AI observability use cases, enabling correlation of model inference data with system telemetry