Maple

About Maple

Maple is open-source observability — traces, logs and metrics over OpenTelemetry, stored in ClickHouse — built around the claim that the incident workflow should not require stitching four tools together. Every signal shares one trace id, so an alert arrives carrying the service, the threshold it broke and sample traces rather than a number and a shrug, and from there you move between the span tree, the structured logs, the browser session replay and the metrics without changing tools or losing the thread. Session replay is genuinely integrated rather than bolted on: every click, route change, console line and failed request is captured, and the replay and the spans share a session id, so a user-reported bug resolves to the exact backend trace. The AI surface is native rather than decorative — an MCP server ships on the plan, alongside AI chat and AI error triage, so an agent can query billions of rows itself instead of a human writing the query. The pricing model is the sharpest thing about it: usage-based with no per-host, no per-seat and no per-query charge, which the cost calculator on the site uses to argue an 85% saving against Datadog at a representative volume. Everything is on GitHub, and Maple Local ships the whole platform as a single binary. Two constraints: retention is 30 days on the Startup plan, and a card is required to start the 14-day trial.

Key Features

Tags