Introduction

Sazabi is an AI-native observability platform. Connect your logs and ask the Sazabi agent about your systems in plain language, or let it detect issues and bring you the answer.

Connect your logs, then ask the Sazabi agent what is happening in production. It searches your data, correlates it, and answers in plain language with the evidence it found. There are no query languages to learn and no dashboards to build.

Quickstart

Create a project, send your first log, and ask your first question in about 10 minutes.

Connect your coding agent

Investigate production from Claude Code, Cursor, Codex, and other agents.

Setup checklist

The steps that take a new project to a working investigation surface.

Concepts

The day-zero ideas: threads, the agent, log sources, issues, and integrations.

What Sazabi does

Most observability tools make you work for answers: know which dashboard to open, write the query, then correlate data across views. Sazabi inverts that. You describe the problem, and the Sazabi agent runs the queries, reads the results, and reports back.

Why are checkout errors elevated right now?
What changed in the last hour that could explain this latency spike?
Show me the error logs for order 8842 from today.

Sazabi also watches your systems on its own. When it detects a problem worth tracking, it opens an issue and can notify your team — with the impact, the likely cause, and the evidence attached, not a raw threshold alert.

How it works

Connect your data

Send logs from any language, framework, or cloud — through OpenTelemetry, a native SDK, or a direct connection to platforms such as AWS CloudWatch, Google Cloud, and Vercel. Sazabi stores and indexes your logs in managed storage so they are ready to search.

Get alerted, or ask

Sazabi detects issues in your data and notifies you through the channels you configure. You can also open a thread any time and ask a question in plain language.

Get answers with evidence

The Sazabi agent responds with what it found — specific log lines, error patterns, and timing correlations — and shows the queries it ran so you can verify and dig deeper.