OllyChat: AI Agents for DevOps

OllyChat allows you to create custom DevOps AI agents that understand and manage your infrastructure.

Engineers are drowning in reliability problems

According to a recent New Relic report:

• Companies have 97 minutes of downtime every week - each minute bleeding $15,000 from your bottom line
• ~400 minutes of outages per month - keeping engineers in reactive firefighting mode and not innovating
• 90% of incident response time is wasted on “ClickOps"
• hunting through dashboards instead of solving problems.

The future of observability isn’t more dashboards and alerts. It’s purpose-built AI agents that work like an expert SRE (site reliability engineer) — without the burnout.

Meet OllyChat

OllyChat isn't another tool – it's an AI teammate that works the way your team works:

• Embedded in your workflow - Slack, Teams, JIRA – wherever your teams communicate
• Understands your infrastructure - Seamlessly integrated with your SRE toolchain
• AI-powered automation - Resolves incidents faster than any human could

The AI-First Open Source DevOps Platform

Today's observability and incident management tools weren't built for AI. They were built for humans – limited by dashboards, on-call response times, and manual processes.

OllyChat flips the script. It's an AI-first platform that:

• Diagnoses incidents in real-time
• Automates remediation with AI-driven agents you control
• Learns from your unique infrastructure and workflows over time

Get Started With Our Demo Prometheus Server

You can try OllyChat without any observability data by pointing your config at our demo Prometheus server: http://34.123.158.139:9090

✨ Features

• Chat with your observability data
• Create your own DevOps AI Agent

🖥️ OllyChat in CLI

💬 OllyChat in Slack

📊 Supported Observability Platforms

• Prometheus
• New Relic (coming soon)
• DataDog (coming soon)
• Elastic (coming soon)

🧠 Supported LLM's

• OpenAI
• Anthropic
• Ollama (tool supporting)

🚀 Getting Started Without Docker

Prerequisites

• node.js
• npm

Install

# Clone the repository
git clone https://github.com/alexkroman/ollychat.git

cd ollychat

# Install dependencies
npm install

Set Your Environment Variables

cp env-example.sh .env
Edit `.env` with your settings

Running the CLI

# Start the CLI
npm run cli:start

Installing Optional Slack Integration

1. Create a Slack App:

2. Visit Slack API

3. Click "Create New App" → "From an app manifest"
4. Select your workspace

5. Copy JSON in deploy/slack/slack-manifest.json

6. Install the app to your workspace:

7. Navigate to "Install App" in your Slack App settings

8. Click "Install to Workspace"

9. Grant requested permissions

10. Start the Slack bot:

# Start the Slack backend
npm run slack:start

Using OllyChat in Slack

Invite @olly to your team or incident channel.

• @olly what's the health status of my cluster?
• @olly any alerts?
• @olly which pods are consuming the most memory?

Getting Started With Docker

You can use Docker to deploy both the Slack app and CLI app

Set Your Docker Environment Variables

cp env-example.sh .env
# Edit .env with your settings

Build the Container

# Build the docker container
npm run docker:build

Run the CLI

# Build the docker container
npm run docker:run:cli

Run the Slack App

# Build the docker container
npm run docker:run:slack

❤️ Contributing

• Submit a feature request or bug report

Building your own DevOps AI Agent with OllyChat

• You can use src/tools/prometheus.ts as a template to get started

# Use prometheus agent as a template
cp src/tools/prometheus.ts src/tools/newTool.ts

• Add your tool as an import and to the tools array in src/tools/index.ts

Run Tests

npm test

Run Linter

npm run lint

Generate Evaluation Run

When you make changes to the app you should generate an evaluation run to test your change against ground truths.

# Load the latest version of the evaluation data
npm run evals:load

# Run the evaluations
npm run evals:start

📜 License

Ollychat is MIT licensed. See LICENSE for details.