Documentation

Quick Start

Install the ReductrAI agent with a single command:

# Install (Linux/macOS)
curl -sSL https://reductrai.com/install | sh

# Set your license key
export REDUCTRAI_LICENSE=your-license-key

# Start the agent
reductrai start

# Point your telemetry to
http://localhost:8080

Platforms: Linux (AMD64), macOS (Intel & Apple Silicon), Windows (AMD64)

Getting Started

Installation, configuration, and first steps

Architecture

How the hybrid SaaS model works

API Reference

REST API endpoints and examples

Pricing

Tiers, limits, and overage billing

Integrations

50+ supported telemetry formats

Troubleshooting

Common issues and solutions

AI Queries

How to ask questions about your infrastructure

Kubernetes

Deploy in Kubernetes environments

How It Works

1. Install Locally

Single Go binary runs on your infrastructure. Your raw telemetry never leaves your machine.

2. Auto-Detect Telemetry

Point your existing monitoring tools at port 8080. We auto-detect 50+ formats including Datadog, Prometheus, OTLP, and more.

3. Anomaly Detection

Local ML detects error spikes, latency anomalies, and traffic drops. Only ~2KB summaries go to cloud.

4. AI Investigation

Cloud LLM analyzes summaries, traces dependencies, and identifies root cause (~$0.03/investigation).

5. Human Approval L3

You receive a notification with root cause and remediation options. Approve with one tap.

Pricing

Tier	Price	Investigations	Overage
Free	$0	10/mo	Blocked
Pro	$149/mo	50/mo	$0.10/ea
Business	$499/mo	200/mo	$0.05/ea
Enterprise	Custom	500+/mo	$0.05/ea

Annual billing saves 20%. BYOK (Bring Your Own Key) available on paid plans for unlimited investigations.

Autonomy Levels

L1 Detect: AI alerts only. You investigate and fix.
L2 Investigate: AI finds root cause. You decide and execute.
L3 Approve: AI proposes fix. You approve or reject. Default
L4 Auto: AI executes known patterns automatically. (Business+)

CLI Commands

reductrai start      # Start the agent
reductrai stop       # Stop the agent
reductrai status     # Check status
reductrai schema     # Show database tables and example queries
reductrai query SQL  # Run local DuckDB query
reductrai version    # Show version
reductrai help       # Show help

Database Schema

Your telemetry is stored in three tables. Run reductrai schema for full details.

spans (traces)

timestamp, trace_id, span_id, parent_span_id, service, operation, duration_us, status, tags

logs

timestamp, service, level, message, tags

metrics

timestamp, service, name, value, tags

Example Queries

# Error rate by service (last hour)
reductrai query "SELECT service, COUNT(*) as total,
  SUM(CASE WHEN status='ERROR' THEN 1 ELSE 0 END) as errors
  FROM spans WHERE timestamp > now() - interval '1 hour'
  GROUP BY service"

# Top 10 slowest operations
reductrai query "SELECT service, operation,
  AVG(duration_us)/1000 as avg_ms
  FROM spans GROUP BY service, operation
  ORDER BY avg_ms DESC LIMIT 10"

# Recent error logs
reductrai query "SELECT timestamp, service, message
  FROM logs WHERE level = 'ERROR'
  ORDER BY timestamp DESC LIMIT 20"

All local queries are FREE and unlimited.

Support

Need help?