← All companies

Applications / Company profile

Augment Code

What Augment Code does

Augment Code (brand/product name: Augment Code) is a B2B AI software-development platform aimed at engineering teams that need code assistance grounded in their actual repositories and workflows, not generic file-by-file suggestions. The company’s core positioning is “codebase understanding” at enterprise scale, exposed through products that support both interactive assistance and agentic workflows—especially around code review and broader portions of the SDLC. Augment Code’s “Context Engine” provides semantic access to a user’s repository so coding agents can retrieve architecture, patterns, conventions, and relevant code context; it is also offered via Model Context Protocol (MCP) so other MCP-compatible agents can use it.

On top of that context layer, Augment Code’s platform includes (1) Cosmos, which it describes as an orchestration “operating system” for coordinating software-development “loops” across the SDLC, including code review, incident response, and other experts; (2) Auggie CLI, a terminal-based agent that runs plans and tool calls in interactive or automation modes; and (3) Augment Code Review, an AI code review agent positioned for large, long-lived codebases.

Technically, Augment Code emphasizes retrieval/indexing and agent orchestration so teams can delegate repetitive work (authoring, review, verification, automation triggers) to agents while keeping human “quality gate” checkpoints. In parallel, the company is continuing to broaden integrations and routing strategies—for example by launching Context Engine MCP for MCP-compatible agents and adding multi-model routing (“Prism”) to reduce cost while preserving quality on its internal benchmarks.

Strategically, Augment Code appears to be building toward AI-native team workflows: it is expanding agent orchestration (Cosmos), tightening review/verification coverage (Code Review and related automation), and integrating security tooling (partnership with Snyk) so agent-generated changes can be checked and remediated inside development pipelines.

News

Company record

May 02, 2026 · Official · Augment CodeIntroducing Augment Prism: model routing to reduce cost and maintain quality

Augment introduced Prism as a new model-picker option that routes each user turn to the underlying model best fitting the work, with claimed savings of up to ~30% lower cost per month for a team sending 10,000 requests/month.

Apr 07, 2026 · Official · Augment CodeThe era of single-model engineering is over

Augment announced adding Gemini 3.1 Pro as its third model alongside Claude and GPT-5, framing the update as part of moving away from single-model engineering.

Feb 06, 2026 · Official · Augment CodeAugment’s Context Engine is now available for any AI coding agent

Augment launched “Context Engine MCP” to bring its semantic codebase retrieval to MCP-compatible agents; the post includes benchmark claims that adding Context Engine improved agent performance by 70%+ across Claude Code, Cursor, and Codex and notes 1,000 free requests for Augment Code users in February.

Jan 09, 2026 · Official · Augment CodeAutomate security at scale: How Augment Code and Snyk eliminate manual security bottlenecks

Augment announced a partnership with Snyk that embeds Snyk’s security intelligence into Auggie CLI so teams can orchestrate security checks, vulnerability detection, and automated remediation within development pipelines.

Dec 11, 2025 · Official · Augment CodeCode review at scale is broken. Here's how we're fixing it.

Augment launched Augment Code Review, describing it as a GPT-5.2-powered review agent for large, long-lived codebases; the post includes competitive benchmarking claims and states the agent is now available to all Augment Code users with a one-week free period for some plans and free access for open source projects via request.

Date not disclosed · Official · Augment CodeAugment Code achieves SOC2 Type II

Augment Code published a post stating it achieved SOC 2 Type II and describes how security and privacy principles are built into the platform.

Source map · 2 recurring channels · 9 references

Still resolving: Newsroom · Official X

What it builds

Cosmos

Cosmos is Augment’s orchestration layer for running AI software-development “loops” across the SDLC, coordinating agents, tools, policies, and memory at an organizational level.

source ↗
Context Engine

Context Engine is Augment’s repository-scale semantic retrieval layer that provides codebase context to agents; it is offered via MCP as “Context Engine MCP” for MCP-compatible agents.

source ↗
Context Engine MCP

Context Engine MCP exposes Augment’s semantic search as an MCP server so any MCP-compatible AI coding agent can retrieve deep codebase context in local or remote mode.

source ↗
Auggie CLI

Auggie CLI is a terminal-based agent that uses Augment’s context and agentic capabilities to plan, execute, and review work across large codebases, with interactive and automation/CI modes.

source ↗
Augment Code Review

Augment Code Review is an AI code review agent designed for large, long-lived codebases, positioned to catch correctness, architectural, and cross-system issues while reducing review noise.

source ↗
Augment Prism (model routing)

Augment Prism is an option in Augment’s model picker that routes user turns to underlying models best fitting each task, aiming to reduce cost while keeping quality near that of routed-to frontier reasoning models.

source ↗

Key metrics

Milestones & partnerships