Augment Code

What Augment Code does

Augment Code provides Cosmos, an agent orchestration platform that enables agentic software development at organizational scale. It helps engineering teams automate various stages of the software development lifecycle (SDLC) by orchestrating AI agents. The platform aims to increase engineering throughput, accelerate time to merge, and reduce token costs. Cosmos offers capabilities like routing tasks to the best models, using a Context Engine for efficient codebase understanding, and enabling shared memory and reusable experts across teams. It integrates with existing developer tools and infrastructure, supporting various deployment options. Augment Code also offers Auggie CLI for interactive and automated agent execution in the terminal, and specialized solutions for code review, incident management, and test coverage.

Products

• CosmosAn agent orchestration platform that runs software agents at scale, providing context, tools, and feedback loops for improved workflows across the SDLC. It includes features like Prism for model routing, a Context Engine for codebase understanding, and an Expert Registry for sharing agents.
• Auggie CLIA command-line interface tool that allows developers to run context-aware agents in their terminal for tasks like code review, test failure investigation, and incident response. It supports interactive and automation modes, and integrates with CI/CD pipelines.
• Context Engine MCPA plug-in that provides semantic code search and codebase understanding to MCP-compatible coding agents, improving code quality and task completion speed while reducing token usage. It offers real-time indexing and multi-source context capabilities.

Key capabilities

• Agent orchestration
• Model routing (Prism)
• Context Engine for codebase understanding
• Expert Registry for agent discovery and sharing
• Human-in-the-Loop escalation
• Integrations with Slack, GitHub, Jira, CI
• Organization Knowledge for shared memory
• Agent Runtime for scheduling and isolation
• Trigger & Automation for SDLC events
• Shared File System
• Sandboxes for isolated execution
• Works on laptops, dev VMs, codespaces, devcontainers, Augment Cloud, and customer's cloud
• Interactive and automation modes for CLI
• Resumable sessions in CLI
• Prompt Enhancer
• User & Repo Rules
• Parallel Agents
• Headless Mode for CI
• Session Sharing
• Streaming Output
• MCP Support
• Task Manager
• Custom Commands
• Tool Permissions
• Multi-model support (Claude, GPT-4, bring your own)
• Proof-of-Possession API
• Non-extractable architecture
• Customer control over data
• Built-in dashboards for usage and trends
• Advanced Analytics for Code Review (Enterprise)
• User Allowlists for Code Review (Enterprise)
• MCP Configuration for Code Review (Enterprise)
• Multi-org Support for Code Review (Enterprise)
• Unlimited Seats & Repos for Code Review (Enterprise)
• Real-time indexing for Context Engine MCP
• Multi-source context for Context Engine MCP
• Customizable incident management workflows
• Customizable delegation workflows

Use cases

• Orchestrating agents across the SDLC
• Deep code review
• CI failure investigation
• PR review enforcement
• Security vulnerability triaging
• Dependency upgrade automation
• Incident response coordination
• Automating code generation
• Automating code review
• Automating test coverage
• Automating issue triage
• Automating incident response
• Automating ticket to PR workflow
• Building features with agents
• Debugging issues with agents
• Exploring codebases with agents
• Automating dependency updates
• Automating regression and E2E testing
• CI health and flake elimination
• Legacy coverage retrofit

AI approach

Augment Code provides an agent orchestration platform called Cosmos that uses AI agents to automate and streamline the software development lifecycle. It leverages a Context Engine for deep codebase understanding and routes tasks to various LLMs (Claude, Gemini, GPT-4) based on cost and quality. The platform focuses on coordinating agents, providing context, and enabling feedback loops to improve performance. It also offers an AI coding agent for the terminal (Auggie CLI) and an MCP-compatible Context Engine for other coding agents.

Tech named: Claude Sonnet 4.6, Claude Opus 4, Gemini 2.5 Pro, GPT-4, LLM, Context Engine

What it says sets it apart

• Agent orchestration at organizational scale
• Prism routes each turn to the right model for cost and quality optimization
• Context Engine maps codebase structurally for efficient token usage
• Shared memory and reusable experts for organizational learning
• Agents own the whole loop, not just the code
• Ships with experts for every stage of the SDLC
• Code review optimized for recall for agents, not precision for humans
• Incident management drives repetitive investigation, pulling humans in for judgment
• Ticket to PR workflow delegates entire tasks to agents, returning reviewed PRs
• Uncompromising data protection with no training on customer data
• Proof-of-Possession API for secure code completions
• Non-extractable architecture prevents data exfiltration
• ISO/IEC 42001-certified Artificial Intelligence Management System
• Scales from startups to Fortune 500
• Context Engine MCP improves agent quality by 70%+
• Context Engine MCP makes agents 2x faster and uses fewer tokens

Funding rounds we track

From the AI funding tracker — rounds as reported by the linked publications.