Amplifier¶

Amplify human capability through AI partnership

Amplifier is an experimental platform that multiplies what you can explore and build by making AI assistants dramatically more effective through domain knowledge, context, and modular capabilities.

Quick Start¶

# Install
uv tool install git+https://github.com/microsoft/amplifier

# Configure
amplifier init

# Run
amplifier

Stay Updated

Amplifier is in active development. Run amplifier update daily to get the latest features and fixes.

Get Started → View on GitHub →

The Challenge¶

"I have more ideas than time to try them out."

Modern AI tools are powerful, but they start from zero every time. They lack your domain knowledge, patterns, context from previous work, and ability to explore multiple approaches simultaneously.

How Amplifier Helps¶

Amplifier creates a supercharged environment where AI assistants become dramatically more effective:

Knowledge & Context

Provides AI with your domain knowledge, patterns, and previous work. Every interaction builds on what came before.

Parallel Exploration

Test multiple approaches simultaneously. Where you might explore 1-2 solutions, Amplifier enables exploring 20 in parallel.

Modular Capabilities

Swap providers, tools, and execution strategies. Build custom workflows that fit your needs.

Compounding Progress

Each capability you add makes the system more capable of building the next. Fast iteration, learning, and evolution.

The Platform¶

Modular architecture inspired by the Linux kernel model:

Modular by Design

Swap providers (Anthropic, OpenAI, Azure, Ollama), tools, and orchestration strategies without changing your code.

Kernel Philosophy

Tiny, stable core (~2,600 lines) that rarely changes. All features live at the edges as replaceable modules.

Agent Delegation

Spawn specialized sub-agents for focused tasks. Each agent has its own tools, context, and capabilities.

Profile System

Pre-configured capability sets from minimal to full-featured. Create your own profiles for repeatable workflows.

Architecture Overview¶

┌─────────────────────────────────────────────────────────────┐
│  amplifier-core (KERNEL) - ~2,600 lines                     │
│  • Ultra-thin, stable, boring                               │
│  • Mechanisms ONLY (loading, coordinating, events)          │
│  • NEVER decides policy                                     │
└─────────────────────────────────────────────────────────────┘
                             ▲
                             │ stable contracts
                             ▼
┌─────────────────────────────────────────────────────────────┐
│  MODULES (Userspace - Swappable)                            │
│  • Providers: LLM backends (Anthropic, OpenAI, Azure, etc.) │
│  • Tools: Capabilities (filesystem, bash, web, search)      │
│  • Orchestrators: Execution loops (basic, streaming, events)│
│  • Contexts: Memory management (simple, persistent)         │
│  • Hooks: Observability (logging, approval, redaction)      │
└─────────────────────────────────────────────────────────────┘

The center stays still so the edges can move fast.

Learn More →

Ecosystem¶

Amplifier provides a rich ecosystem of swappable modules and libraries.

Type	Examples	Purpose
Providers	Anthropic, OpenAI, Azure, Ollama	LLM backend integrations
Tools	Filesystem, Bash, Web, Search	Agent capabilities
Orchestrators	Basic, Streaming, Events	Execution loop strategies
Contexts	Simple, Persistent	Conversation memory
Hooks	Logging, Approval, Redaction	Observability & control
Libraries	Profiles, Collections, Config	App-layer functionality

Browse Ecosystem → Community Showcase →

Target Outcomes¶

We're discovering what's possible when AI partnership amplifies human capability:

Ideas that take weeks → hours - Rapid exploration of possibilities
Test 10x more approaches - Parallel experimentation at scale
Knowledge compounds - Every project accelerates the next
Time from idea to prototype → near zero - Fast iteration and learning

This is an experimental journey. We're building Amplifier with Amplifier, progressively discovering what amplification makes possible.

Who Is This For?¶

Explorers

Fork and build your own amplification systems. Share discoveries and push boundaries of what's possible.

Teams

Share profiles and collections across your organization. Consistent AI workflows for everyone.

Developers

Build AI-powered development workflows. Code review, refactoring, debugging, and documentation assistance.

Module Authors

Create custom providers, tools, or hooks. Stable contracts make modules independently developable.

Next Steps¶

Installation - Get Amplifier running
Getting Started - Your first AI session
CLI Reference - All available commands
Architecture - How it all fits together
Module Development - Build your own modules

Community¶

Amplifier is a Microsoft open source project.