Anaconda Acquires Kilo Code to Expand Into Agentic Software Development

Anaconda has acquired Kilo Code, an open-source, model-agnostic platform that embeds AI agents into the tools developers use to build software. The deal brings Kilo’s community of more than three million developers into Anaconda’s expanding enterprise AI platform, extending the company’s reach from Python packages and production orchestration into integrated development environments and command-line workflows. The announcement did not disclose financial terms.

Kilo will continue operating without immediate changes to its existing products, pricing plans or customer support. Anaconda said additional details about how the two platforms will be integrated will be announced later.

Bringing AI Agents Into the Development Environment

Kilo Code is designed for what the company calls “agentic engineering,” where AI systems do more than suggest individual lines of code. Its agents can examine repositories, plan features, modify files, investigate errors and execute multi-step development tasks.

The platform works across Visual Studio Code, JetBrains integrated development environments, a command-line interface and cloud-based agents. It provides several operating modes, including a coding mode for writing and refactoring software, an architecture mode for planning complex projects and a debugging mode that examines errors and traces potential problems through a codebase. Kilo has also expanded into automated code review, remote agents and integrations with collaboration tools such as Slack.

This gives Anaconda a direct presence at the point where developers first interact with AI while building an application. Rather than entering the workflow primarily when teams assemble Python environments or prepare workloads for deployment, Anaconda will now have technology embedded inside the environments where code and AI-generated artifacts are created.

A Model-Agnostic Approach to AI Coding

One of Kilo’s more consequential features is its refusal to center the platform around a single AI model provider. Its gateway provides access to more than 500 models through a unified interface, covering commercial frontier systems, open-weight models and specialized alternatives.

Developers can select models directly or use automatic routing to match different models with different workloads. Kilo’s routing options include strategies focused on model capability, lower costs, balanced performance or free models. The platform can therefore assign a more capable model to complex architectural work while using a less expensive model for routine engineering tasks.

Kilo also supports multiple development surfaces from the same platform, including local integrated development environments, mobile applications, cloud agents and managed agent deployments. This broader scope distinguishes it from coding assistants that remain limited to autocomplete or chat inside a single editor.

Model flexibility is becoming increasingly relevant as organizations consume larger volumes of tokens through software development agents. Kilo says it now orchestrates almost 10 trillion tokens per month. At that scale, the ability to route requests based on task complexity, price and organizational policy can have a substantial effect on costs and operational dependencies.

Extending Anaconda’s Enterprise AI Strategy

The acquisition is the latest step in Anaconda’s effort to reposition itself as a broader foundation for AI-native software development.

Anaconda is best known for its Python distribution and package management ecosystem, which helps developers install libraries, resolve dependencies and create reproducible environments. Its enterprise platform adds controls for validating packages, scanning for security vulnerabilities and managing the open-source components that enter development environments.

The company has increasingly moved beyond package management. The Anaconda Platform now combines its Python foundation with governed environments and tools for building and operating production AI workflows. Its AI orchestration layer provides workflow execution, observability, traceability and infrastructure support intended to help teams move models from experimentation into deployed applications.

Kilo fills another part of that lifecycle. Anaconda can now connect the development environment where an agent receives its first prompt with the packages, models, infrastructure and orchestration systems used to run the resulting workload.

Building on the Outerbounds Acquisition

The Kilo deal follows Anaconda’s April 2026 acquisition of Outerbounds, the company behind the open-source Metaflow framework originally created at Netflix.

Outerbounds added production workflow orchestration, artifact tracking, experiment management and scalable compute to the Anaconda Platform. Its technology is designed to let data scientists and machine learning engineers move workloads across cloud, data-platform and hybrid infrastructure without forcing them to adopt an entirely new development model.

With Outerbounds, Anaconda gained more of the infrastructure required to execute and observe AI systems in production. Kilo adds the developer-facing agent layer at the other end of the process. Together, the acquisitions give Anaconda technology spanning the initial AI-assisted coding session, package and model selection, environment management, workflow orchestration and production deployment.

That does not mean all of these components are already operating as a single product. Anaconda has explicitly described the deeper connection between Kilo and its governed packages, models and environments as a direction it is building toward, rather than a fully available capability.

Governance Moves Closer to the First Prompt

The enterprise rationale for the acquisition is largely centered on visibility and control.

As developers adopt AI coding agents through personal accounts, separate application programming interface keys and multiple model providers, companies can struggle to determine which models are being used, where proprietary code is being sent and how much individual teams are spending.

Anaconda’s existing security tools scan packages, models and dependencies before they enter production environments. The platform also supports policy enforcement, audit trails, vulnerability monitoring and the generation of software and AI bills of materials that document the components included in a project.

Connecting these controls with Kilo could eventually allow organizations to apply model and package policies much earlier in development. A company could, for example, restrict access to unapproved models, maintain records of AI-assisted development activity or direct sensitive workloads toward self-hosted systems.

The practical value will depend on how deeply the platforms are integrated. Enterprise governance can easily become another layer of friction when policies are poorly implemented, particularly if controls interfere with the flexibility that attracted developers to Kilo in the first place.

What the Acquisition Means for Kilo Users

For existing Kilo users, little changes immediately. The coding agent remains available across VS Code, JetBrains, the command line and cloud environments, with continued support for multiple models and providers.

The acquisition may give Kilo access to Anaconda’s enterprise distribution, security expertise and relationships with large organizations. Anaconda says its ecosystem reaches more than 52 million users and is used by 95% of the Fortune 500, giving Kilo a path into companies that may have been reluctant to approve standalone AI development tools.

At the same time, Anaconda will need to preserve the qualities that drove Kilo’s adoption, particularly its open-source roots and model-agnostic architecture. An enterprise integration that limits model choice or makes the product more difficult for individual developers to use would undermine a central part of the acquisition’s strategic logic.

Future Implications

The acquisition reflects a broader consolidation taking place across AI software development. Coding agents, model gateways, package security, workflow orchestration and production governance have generally emerged as separate product categories. Anaconda is attempting to bring those layers together under a common platform.

Its strategy now covers three major stages of AI-native development: Kilo provides the agentic engineering environment, Anaconda Core manages packages and reproducible environments, and Outerbounds supplies production orchestration.

The opportunity is to create a more continuous path from an initial prompt to a governed production deployment. The challenge will be turning a collection of acquired and existing technologies into a coherent platform without reducing developer choice or introducing excessive complexity.

Kilo’s products will remain available independently while that integration proceeds. The longer-term importance of the acquisition will therefore depend less on the number of supported models or tokens currently processed, and more on whether Anaconda can connect developer-friendly AI agents with enterprise security and production infrastructure without compromising either side.