Talks
What’s on stage at CPH^Automate, with speaker details. Timeslots are placeholders until the schedule is finalized closer to the event.
14:15 – 14:45
We asked two route planners to write down how they route fibre and got back 98 sections — which turned out to describe routing by *who owns the building*, while the same engineers had for years actually been routing by *what's physically installed there*. This is how we stopped asking people to remember, recovered the real rules by replaying historical decisions out of the inventory database, why the target turned out to be agreement rather than correctness, and the one line of code I invented by accident that got one in four trunk selections wrong without ever raising an error.
10:15 – 10:45
Transformation Through Community
Andreas Bækdahl, Emil Commerou
We turned a classic professional services organization into a curious automation-forward team. Network automation does not scale through tools and platform alone — we built a thriving network automation community in Wingmen Solutions through 6 individual threads:
- Platform: Remove developer friction.
- Learning: A curious space.
- Documentation: Tailored recipes for success.
- Support: Open door policy.
- Lab: Break while you make.
- Tools: Aim for good enough; just in time.
11:45 – 12:15
JSON freedom or chaos; how to trust your data
Can you trust your data?
In this talk we'll take a real life use case where the flexibility desires and time have gotten the best of it. What happens when you deal with "freedom" data? How to go from this to a better future?
It is a technical talk how we recognise changes are desired, validation is required for trust to grow. Pydantic and pytest are used to work towards a solution where failing fast is a main objective.
11:15 – 11:45
Bootstrapping a Source of Truth in a Brownfield Environment: Field-Tested Patterns
Going through the challenges of initializing a SoT in an already existing environment, a few design and implementation patterns to solve them (consolidating a by-spine with a few basic fields, aligning all referentials on this and then iterating on it, for example). Common mistakes and finally how to maintain it over time.
It is based on my experience as a network automation consultant who designed and implemented a dozen SoTs (mostly NetBox and a few Infrahub instances) in diverse enterprise environments.
09:15 – 09:45
What Should You Build When You Can Build Anything?
Being able to build almost anything does not make decisions easier. It creates more possibilities, and therefore more difficult choices. Every time we ask, "Should we build this?", we must also consider a "buy" option that AI is making better and faster.
Most discussions focus on how AI helps individuals build their own tools. But we speak less about what AI means for the vendors competing with these tools. This is also an important part of the build-versus-buy decision. AI is changing the capabilities of both internal teams and software vendors, so the comparison between them is changing as well.
In this talk, we will take a step back from the hype and consider what the software industry may look like after this transition. What will still be worth building ourselves? What will be better to buy? And what will be assembled from existing components?
We will explore these questions in the specific context of network automation. What do these changes mean for our industry, and how will AI affect what network automation teams choose to build, buy, or assemble from existing components?
The talk will not provide one answer that works for everyone.
Instead, it will offer a framework—and better questions—to help you decide what to build, what to buy, and what to combine.
09:45 – 10:15
Democratizing Network Automation: What We Learned Building an Open Automation Stack
Network automation is often presented as something reserved for the largest operators with dedicated automation teams and substantial budgets. This presentation tells the story of how a regional operator and Ductus took a different approach: focusing on a solid architectural foundation, an incremental delivery model, and open technologies to build a sustainable automation platform. The result isn't just a successful implementation—it's a demonstration that advanced network automation can be achievable for operators of all sizes.
12:15 – 12:45
Keeping ~2k IP Pools Filled with Just-in-Time Provisioning
In a constantly changing network, efficient utilization of scarce IPv4 address space is critical. Learn how we closed the automation loop by leveraging NETCONF notification streaming to dynamically provision IP pools exactly when and where they are needed.
13:45 – 14:15
Vibe-ops: Migrating production infrastructure with an AI agent
I migrated our whole infrastructure off VMware in two weeks with Claude Code as my pair-operator — Proxmox, an HA firewall pair, SSO, CI, monitoring, the lot.
This is the honest story of what an AI can actually do in production ops, and the afternoon one firewall change took the entire network down with us outside.
14:45 – 15:15
When Your Source of Truth Isn’t Really Yours: Lessons from Cross-Team Integrations
What happens when another team integrates directly with your Source of Truth and their needs start limiting your ability to evolve your platform?
In this talk, I will share lessons learned from operating a Source of Truth that supported critical business processes while also serving as a dependency for another team through a complex, poorly documented integration. Because the integration was tightly coupled and the consuming team had different organizational priorities, even small improvements to data models, processes, and workflows became difficult to implement.
The session explores the challenges that arise when ownership, responsibilities, and priorities are not clearly separated between teams, and how these issues can slow down innovation and increase operational risk.
I will also discuss practical approaches that helped improve collaboration and reduce complexity, including:
– Creating simpler and more stable interfaces between teams
– Clearly documenting ownership, expectations, and integration contracts
– Implementing automated and continuous testing to detect breaking changes early
– Reducing coupling to enable independent evolution of systems and processes
Attendees will leave with actionable ideas for designing team boundaries and integrations that scale with both technical and organizational change.
16:15 – 16:45
Laziness. A Blessing or a Curse?
For decades we have discussed network automation, continually hoping the next tool or technology will solve it.
Perhaps the issue is not the technology at all. Maybe we are just lazy.
This speech questions a number of unpleasant assumptions concerning automation, AI, engineering culture and our obsession with the next exciting innovation.
This speech presents an uncomfortable examination of the reasons why we could be the main obstacle to network automation.
15:45 – 16:15
Shift Left: Building Network Automation That Lasts
Getting something automated is hard. Keeping it automated as networks, platforms, and business requirements evolve is even harder.
Too often, automation is treated as the final step of a project rather than something that should influence how systems, data models, APIs, and operational processes are designed from the beginning. The result is brittle automation that requires constant maintenance, struggles to adapt to change, and eventually becomes more expensive to maintain than the manual processes it replaced.
A shift-left approach encourages us to think about automation much earlier in the lifecycle. By considering automation during the design of services, APIs, data models, workflows, and operational practices, we can build systems that are easier to automate, easier to evolve, and easier to operate over the long term.
In this session we share ideas and experiences that help move network automation from a collection of scripts to a sustainable engineering discipline.