If there was ever a doubt that network automation has graduated from a niche engineering hobby to a critical business imperative, AutoCon 4 put it to rest last week. From November 17-21, the Network Automation Forum (NAF) descended on the Austin Marriott Downtown in Texas, bringing together the brightest minds in the industry for a week of intense workshops, candid discussions, and a clear look at the future of our field.
As attendees filtered out of the closing ceremony on Friday, the overwhelming sentiment was one of maturation. We’ve moved past the early days of “Hello World” in Python and are now grappling with complex orchestration, agentic AI, and the organizational leadership required to scale these solutions. If you weren’t able to make it to Austin, here is a deep dive into what defined AutoCon 4.
The Evolution of the Agenda: Growing Up
One of the most significant changes this year was the structure of the conference itself. In previous iterations, NAF stuck strictly to a single-track format, ensuring everyone saw the same content. This year, they introduced a bold change on Thursday by splitting the day into two distinct tracks adding an Advanced and Leadership breakout.
This bifurcation perfectly mirrors the state of the industry. On one hand, we have the “Advanced” practitioners who are pushing the boundaries of technology—discussing things like the Model Context Protocol (MCP) and modular Ansible frameworks. On the other, we have the “Leadership” track, focused on the equally difficult challenge of culture and funding.
Talks like Alex Henthorn-Iwane’s How to Build a CFO-Worthy Business Case and Eyvonne Sharp’s Beyond the Code: People-Centric Leadership underscored a critical realization: you can have the best code in the world, but if you can’t sell the business value or manage the human change, your automation initiative will stall. Jeff Gray’s opening keynote, Building a Network Automation Business Case that Wins at the Top, set this tone early, challenging engineers to speak the language of finance and executives. Jeff did a nice job of laying out and example of how to build a business case that shows the Return on Investment (ROI) for starting an automation initiative. Alex’s talk in the Leadership track further articulate the need to build this business case.
The AI Explosion and “Agentic Autonomy”
It wouldn’t be a 2025 tech conference without AI, but AutoCon 4 managed to cut through the hype and focus on practical application. The buzzword of the week was undeniably “Agents.”
The closing keynote by Greg Freeman, The NetDevOps Journey: Manual Firefighting to Agentic Autonomy, was a highlight, painting a picture of a future where network operations aren’t just automated but autonomous. This theme rippled through the rest of the content. We saw sessions like Building AI with AI by Senad Palislamovic and John Capobianco’s From CLI to GPT, which demonstrated how Large Language Models are rewriting the rules of engagement for network engineers.
However, the content wasn’t just theoretical. The Battle of the Bots lightning talk by Eric Chou and the workshops on building AI agents provided hands-on reality checks. The industry is clearly trying to figure out where AI fits—is it a co-pilot, a troubleshooter, or a fully autonomous operator? The consensus seems to be that that tools to build fully autonomous agents exists, the more realist usage involves a human double checking things. AI assisting human operations and configuration is very real and some folks are already doing this in production.
Technical Deep Dives into MCP
For the engineers in the room, the technical content was richer than ever. A standout theme was the Model Context Protocol (MCP), which was the subject of both a sold-out workshop and a dedicated talk by William Collins. As automation ecosystems become more fragmented, standardizing how AI and tools perceive context is becoming crucial.
The Source of Truth Showdown: Data Gets Smart
If there was one clear battleground on the vendor floor this year, it was over the Source of Truth (SoT). What used to be a static repository for IP addresses and VLANs has morphed into the dynamic engine room for the entire automation ecosystem. The message from every vendor was consistent: Your AI is only as good as your data.
NetBox Labs continued its aggressive expansion from open-source option to enterprise powerhouse with growth in their SaaS offering. Their big focus at AutoCon 4 was closing the loop between “intended state” and “actual state.” They showcased NetBox Discovery and NetBox Assurance, tools designed to automatically populate the SoT and validate that the network matches the model. By moving into assurance, NetBox is signaling that it wants to be the single pane of glass for both planning and observing the network, tackling the perennial “operational drift” problem head-on.
On the other side of the ring, Network to Code was showing off how Nautobot is evolving into a full-blown platform. The buzz at their booth was around NautobotGPT, an AI-powered assistant that allows engineers to query their network data using natural language. They also pushed Nautobot Cloud, their SaaS offering, which aims to lower the barrier to entry for teams tired of managing their own infrastructure. For Nautobot, the strategy is clear: make the data accessible and actionable, lowering the “tax” of maintaining a source of truth.
Then there was the challenger, OpsMill, which turned heads with Infrahub. They are fundamentally rethinking the architecture of a Source of Truth by moving away from traditional relational tables to a Knowledge Graph. Their workshop highlighted how a graph database can better map the complex, non-linear relationships of modern infrastructure. They also leaned heavily into versioning—giving the database itself “Git-like” capabilities to branch, merge, and rollback data. It’s a geekier, more architectural pitch, but one that resonated with engineers hitting the limits of traditional schemas.
Community and “The Hallway Track”
Despite the packed agenda, the “Hallway Track”—the informal conversations between sessions—remained the heart of AutoCon. The energy in Austin was electric. There is a specific camaraderie in the network automation community; a shared understanding of the pain of legacy CLI and the joy of a successfully deployed workflow.
The Confessions of a CLI Lifer Who Learned to Love Automation by Andy Lapteff resonated with many attendees who are bridging the gap between traditional networking and modern DevOps. These stories of personal transformation are just as important as the technical ones. They remind us that automation is ultimately about making the lives of network engineers better, reducing burnout, and eliminating the “manual firefighting” that Greg Freeman alluded to.
Looking Forward
As I look back on the week, AutoCon 4 felt like a pivotal moment. The question posed on the conference website—“Why haven’t we seen full adoption of network automation, yet?”—was answered in many ways. It’s not a lack of tools; it’s a matter of integration, culture, and trust.
The focus on “Leadership” gave us the soft skills to navigate corporate politics and articulate the ROI of automation initiatives in terms the executive team understands. Many engineers get frustrated that management does not “just get it” but being able to build a business case that explains it in the right terms will really help with adoption. And the deep dive into AI gave us a glimpse of the velocity we can expect in the coming years.
Austin provided the perfect backdrop for this innovation, but the work is far from over. As Scott Robohn noted in his talk Toward Autonomy, we are on a journey. The destination is a network that is self-driving, self-healing, and resilient, but the road there is paved with the incremental improvements we discussed last week.
For those who left Austin inspired, the countdown has already begun for AutoCon 5, returning to Europe in May. Until then, it’s time to take these lessons back to our terminals and boardrooms and keep building.
Ryburn.org 
Leave a comment