New Beginnings, same principles, still pushing for Open Source and Open Standards
Over the last 6 months, I’ve been spending considerable time exploring the land of coding agents, agentic frameworks (to build business agents), and the rapidly evolving landscape of tooling in these spaces. At KubeCon EU, it wasn’t a shock to see how companies are building their own stacks to run, serve, and control how teams interact with a wide range of LLMs (for coding or business agents).
Last Thursday was my last at Diagrid, where I went knee-deep for the last three years into developer tooling, workflow engines, and, more importantly, how to smooth developers' journeys when building complex distributed systems with a constantly evolving set of tools.
Today I am hyped to be joining Dash0 as an Ecosystem Engineer, because observability, and to be more concrete, OpenTelemetry has a big role, not only on the new stacks that companies are building, but also in how developers gain visibility about what their coding agents are doing. It is becoming clear to me that OpenTelemetry is the foundation for understanding agentic systems.
While companies are building new stacks using mature ecosystems such as Kubernetes and OpenTelemetry, new use cases are emerging as developers push more code than ever before, using tools that haven’t been tested at scale. Understanding what is going on is more important than ever.
Unfortunately, to safely deliver all the software generated today into production, understanding our production environments is not enough. We will need to understand: how code is generated?, how that process works for your organization?, how generated code behaves in production environments (is it deterministic or not)?, how do we evaluate non-deterministic errors?, and, more importantly, how to revert changes based on observed errors?.
I strongly believe that to answer these questions, we need standards that span developer tooling and the infrastructure that will run that code in production. Open Telemetry is strategically positioned to lead the way to other standards, such as the OCI (Open Container Initiative), for packaging and distributing knowledge artifacts so we can reuse not only existing infrastructure but also the practices we have evolved over the last 10 years. These standards are relevant to all the phases of the SDLC (Software Development Life Cycle), not only to developers, not only to DevOps, not only to production.
But let’s be honest: none of the mainstream tools being developed today have observability or industry standards (distribution provenance, security, auditability) as top priorities on their roadmaps. Hence, it is up to us, the “consumers”, the organizations, and the communities evaluating these tools to figure out how they map to the problems that we are trying to solve and how we can build production-ready solutions with them.
As always, you will see me out in the open, collaborating with and learning from the industry's smartest people. I look forward to collaborating, building cool stuff, and interacting with communities spanning from developers and developer tooling to infrastructure and SREs.