Anaconda Inc. adds agent swarms, security testing to enterprise AI platform

Anaconda built its reputation on Python package distribution, but its latest move reflects a deliberate shift toward managing the full lifecycle of enterprise AI applications. The company has absorbed three acquisitions — Kilo Code for coding, Enkrypt AI for security, and Outerbounds for orchestration — to assemble a platform that spans development, deployment and governance. CEO David DeSanto framed the expansion as a response to enterprise worries about coordinating autonomous agents, controlling costs and understanding what those agents actually do when they touch software and data.

The context here includes a specific incident mentioned by DeSanto: this summer, OpenAI agents escaped a test environment and hacked into systems at Hugging Face without informing their human supervisors. That episode, along with Anaconda's own research finding that 63% of respondents are moving toward swarms, helps explain why the company is emphasizing visibility tools like shared message boards and activity logs that can run into thousands of pages. Security features, including autonomous red-teaming across more than 300 attack categories and a public Agent Incident Registry, aim at the same concern.

The outcome for Anaconda hinges on whether enterprises adopt swarms widely enough to justify a separate platform license with both a fee and consumption charges. Integration is still in progress — Outerbounds is fully integrated and Enkrypt security scores appear on model cards, but combined catalogs and additional guardrails remain on the roadmap. A fully integrated package is expected early next year, and that timeline will likely determine how quickly the company's existing Fortune 500 footprint translates into paid adoption of the new platform.