Hello, DevNet Community!

Welcome to the November newsletter! With Thanksgiving around the corner in the US, we want to take a moment to say how grateful we are for our global DevNet community - and a special thanks to all our DevNet Experts, past and present, for their ongoing passion and contributions.

This month, we’re excited to share updates ranging from a new Python SDK for Secure Access to AI-driven networking, observability deep dives, open source contributions, and more.

One quick note: we’re migrating our newsletter system to improve your experience. Because of this upgrade, our next issue will arrive in late January or early February 2026. Thanks for your patience — we can’t wait to connect again soon!

What's new in DevNet

VIDEO | DevOps Shop

Network as Code: Solutions & Tools from Cisco CX

Rafael Miller, a Principal Engineer at Cisco and Oleksii Borysenko, discuss the evolution of "Network as Code" (NaC). He explores how NaC has grown from its ACI origins to cover the entire Cisco portfolio (Meraki, VManage, etc.) and even direct-to-device interaction. The session highlights key open-source tools like nac-validate and nac-test that embed pre- and post-change validation directly into your CI/CD pipeline, ensuring a more reliable and secure approach to network automation.

Watch now >

VIDEO | Mostly Human

The Mostly Human Show: Choosing the Right LLM

In Episode 3 of the "Mostly Human Show," Adrian and Shereen explore how to select the right Large Language Model (LLM) for your networking use case. They also provide a framework for making your choice, balancing task complexity, latency needs, and the power of fine-tuning versus Retrieval-Augmented Generation (RAG).

Watch now >

Cisco Open Source News

Cisco AI Defense

The Cisco AI Defense team is actively leading in the open-source security space: they’ve recently published their “SecureBERT2” research and modelling efforts, launched the “MCP Scanner” project to proactively scan for threats across MCP environments, and rolled out an “AI Defense Python SDK” to make it simple for developers and security teams to incorporate AI-powered defence capabilities into their workflows. Check out all of the projects on the team’s GitHub page, CiscoAI Defense, for full context and latest updates.

CNCF News from KubeCon + CloudNativeCon North America

Cisco teams had a great week at KubeCon and we have great news to bring to you. KServe Project Promotion: The KServe project, a standardized AI inference platform on Kubernetes, has been promoted to CNCF incubating status. CNCF Technology Radar: A new tech radar has been published, focusing on AI inference engines, ML orchestration, and agentic AI platforms.State of Cloud Native Development Report: The latest report from CNCF and SlashData is out, finding that the cloud native ecosystem has surged to 15.6 million developers.

Explore more developer resources

SERIES | Agentic Impressions - Shreen Bellamy

Agent Memory Systems: Beyond Context Windows

Context windows are a major bottleneck for AI agents. In this episode of "Agentic Impressions," Shireen Bellamy explains why architecture beats raw power, demonstrating how to build persistent memory for agents using a combination of vector, episodic, and network-specific (SQLite) databases. This approach solves the "token triangle problem" of cost, latency, and limits, enabling agents to remember interactions across sessions and learn over time.

Watch now >

VIDEO | AI is the Future - Oleksii Borysenko

Securing AI Agent-Generated Code with Project CodeGuard

How do you ensure the code your AI assistant writes is secure? The Cisco AI Security team introduces Project CodeGuard, an open-source, model-agnostic framework that embeds security rules directly into your AI coding workflow. See how it works in IDEs like Cursor and CLIs like Claude to actively prevent vulnerabilities (like hard-coded credentials) from being introduced, moving security to the very start of the development process.

Watch now >

SERIES | NetGru - Adrian Iliesiu

Building a Python Network Emulator to Learn Python | Ep. 83

Tired of learning Python with examples about pets and groceries? Andreas and Adrian show us how to learn Python fundamentals by building something a network engineer actually cares about: a simple Layer 2 switch. This session masterfully teaches core concepts like lists, dictionaries, functions, and classes by using them to build a working MAC address table and a familiar command-line interface (CLI).

Watch now >

BLOG

Cisco AI Networking for Developers: What's New

Cisco has introduced major updates to transform how you build, integrate, and automate across Meraki and Catalyst platforms. For developers, this means more reliable automation, safer network changes, and accelerated rollouts. Highlights include an AI Assistant now in the Meraki dashboard, a new Global Overview for unified visibility (coming Q4 2025), and a "Unified Branch" strategy featuring Branch-as-Code and low-code Cisco Workflows.

Read blog >

Announcing the New Cisco Secure Access Python SDK

To make security orchestration and policy management easier, we've launched a new Python SDK for the Cisco Secure Access API. The SDK implements authentication and usage best practices, handles errors, and helps you manage resource deployment and policy changes. The repo includes examples for managing API credentials, destination lists, access policy rules, and more. All of it compatible with your favorite AI coding assistants.

Read blog >

Like the newsletter?

Keep your peers up to date on recent developments in DevNet resources and industry trends.

Want more?

Explore all Cisco developer resources we have to offer, including labs, docs, sample code, sandbox, and more!