From Group Science Project to Enterprise Service: Rethinking OpenTelemetry

Ari Zilka says he counted 23 different observability vendors at KubeCon + CloudNativeCon  North America 2025, and talked with every single one. According to Zilka, first-ever Hortonworks CPO and now founder of MyDecisive.ai, they’re all showing customers essentially the same thing. “Every observability vendor here at KubeCon is showing you a dashboard that has a magical red piece of text on it, like that’s the broken thing,” Zilka told The New Stack. “Only, except in very contrived use cases, it’s not actually the broken thing.” In this episode of The New Stack Makers, self-described “data center and big data guy” Zilka sat down with TNS Founder and Publisher Alex Williams to discuss why observability needs to move from reactive dashboards to proactive operations, and why OpenTelemetry (OTel) adoption has become unsustainably expensive for most organizations. The MTTR Problem Nobody Was Solving Zilka’s frustration comes from direct experience. During his time at New Relic, he spoke with roughly 2,000 CIOs who all asked for the same thing: reduce mean time to resolution (MTTR) for system errors while also cutting costs. The only honest answer he could give, he said, was that such a solution simply did not exist. “By the time you’re broken, it’s too late,” Zilka said. “What you want to do is make sure that the humans aren’t in the loop making the changes, and that you make every production change staring at your telemetry while you make the change. Who better to stare at telemetry and make a change than an AI?” MyDecisive.ai is Zilka’s answer to this challenge. Built on OpenTelemetry, Kubernetes, Prometheus and other cloud native technologies, the system acts as what Zilka calls “a bump in the wire” to intercept telemetry data and use that to automate decisions like rolling back a bad release when error rates spike. When OpenTelemetry Becomes a Team Sport The impetus for creating MyDecisive, Zilka told the Makers audience, was what he observed happening with Otel adoption itself. He shared the story of a major streaming media company that went from using Platform as a Service (PaaS) with New Relic and Datadog for observability to running 10,000 OTel instances in-house, with 40 people dedicated to maintaining their stack. “Everywhere I go that has OpenTelemetry, you have 20- or 30-person teams,” Zilka observed. “If it’s a $100 million company, it’s also got at least two or three dedicated OTel experts.” Platform engineering teams love OTel, he continued, because they finally have software they can program and own. But this also leaves them “locked arm in arm with Dev to get it right.” MyDecisive’s approach decouples that relationship so platform teams can make autonomous decisions with zero need to ask developers to make corresponding changes in the application layer. “We’ve moved from OpenTelemetry as a group science project to OpenTelemetry as an enterprise-hardened service,” Zilka said. Watch the full conversation to hear more about how MyDecisive.ai’s “if this, then that” interface works, why Zilka chose an open source model, and what proactive operations mean for the future of observability. The post From Group Science Project to Enterprise Service: Rethinking OpenTelemetry appeared first on The New Stack.