AI-native time series database

Drop the legacy TSDB.
Leapfrog into the AI era.

TSDB.ai is the most Storage Efficient Time Series Database on the planet with Native AI capabilities. Connect OpenAI or Anthropic Claude to the built-in Vector and MCP servers and chat with your data.

Native integration with LLMs

OpenAI

Anthropic

Gemini

Llama

MMistral

Bedrock

53.33×

Storage compression

80%

Less disk usage than Prometheus

50%

Less disk usage than VictoriaMetrics

100%

Prometheus compatible

How far does your storage go?

Compressed stored size from raw time series input

1 TB of raw time series

Engine	Stored Size
Prometheus	96 GB
VictoriaMetrics	38.4 GB
TSDB.aiBest	19.2 GB

1 PB of raw time series

Engine	Stored Size
Prometheus	96 TB
VictoriaMetrics	38.4 TB
TSDB.aiBest	19.2 TB

Free tool

See what your storage costs would actually be

Plug in your active time series and retention period to compare real-world monthly storage costs across AWS, Google Cloud, and Azure.

Try the storage cost calculator

Included in every plan

Core — Free Forever

Everything you need to observe and understand your systems at any scale.

Built-in

MCP Server

A fully integrated Model Context Protocol server. Any LLM or AI agent can query live time series, trigger anomaly detection, and read forecasts directly — no middleware, no custom glue code.

Native tool calls: query_metrics, get_anomalies, forecast
Streaming results back to the model context window
Works with Claude, GPT-4o, Gemini, and any MCP client

Built-in

Vector Store

Time series and anomaly events are embedded and indexed automatically. Ask semantic questions in plain English and get back exact metric windows — no SQL, no PromQL required.

Embeddings over metric patterns, anomalies, and regime changes
Semantic search: "show me all CPU spikes during deploys"
Powers AI Assistant context without manual retrieval pipelines

Real-time Dashboard

Live metric charts with zoom, pan, and time-range selection

AI Chat

Ask your TSDB instance questions in natural language, grounded in your actual data

Anomaly Detection

Automatic spike, dip, and deviation detection across all ingested metrics

Forecasting

ML-powered time series forecasting with confidence bands

Pattern Recognition

Discovers repeating behavioral patterns — daily cycles, weekly rhythms

Regime Change Detection

Identifies when a metric's statistical baseline permanently shifts

Prometheus-compatible Ingest

Drop-in replacement for any Prometheus-based scraping setup

Scraper Setup Wizard

Auto-generates scraper configs for Linux, macOS, Docker, and Kubernetes

Model Compression Engine

Polynomial-fit compression; stores far less than raw samples with tunable RMSE tolerance

WAL + Block Storage

Durable write-ahead log with staged and canonical block layers

Self-monitoring

TSDB.ai instruments itself and exposes its own metrics endpoint

Live Configuration UI

Tune ingestion, storage, compression, and causal analysis without touching files

Instance Inspector

Real-time view of all internal components and their health status

Pro License — $1,200 / year

Built for teams that can't afford downtime.

Pro adds intelligent alerting, deep root cause analysis, and team chat integrations. One license, unlimited users on your instance.

Alert Builder

Threshold, RMSE, % change, anomaly, or forecast-breach rules — route to Slack, Teams, Telegram, or webhooks

Chat Integrations

Connect Slack, Teams, Webex, and Telegram with per-platform event filters and one-click delivery test

Root Cause Graph

Auto-discovers cause-and-effect between metrics with quantified lag times and confidence scores

Forecast-breach Alerts

Fire alerts before a metric goes out of range based on projected trajectory

Natural-language Rule Preview

See a plain-English description of your alert rule before saving

Drop-in Prometheus replacement

Already using Prometheus? Add two lines to your configuration and start sending data to TSDB.ai. Your existing dashboards, alerts, and PromQL queries continue to work without modification.

Linux

macOS

Docker

Kubernetes

Grafana

Alertmanager

Open Source

Get running in minutes.

Self-host TSDB.ai locally or in Docker. Go 1.23+ and Node.js 18+ are the only dependencies.

git clone https://github.com/tsdbai/tsdb.ai
cd tsdb.ai/v0.9

# Start all services
./install/local/start-server.sh

# In a new terminal — start the admin UI
./install/local/start-ui.sh

# UI → http://localhost:3000

Then open http://localhost:3000Full installation docs →

Start observing in minutes.

Free tier is free forever. No credit card. No limits on metrics.

Drop the legacy TSDB.Leapfrog into the AI era.