Building the Platform
That Snowflake Couldn't Resist
The call came in. The system was down. The team opened five different tools - one for logs, one for metrics, one for traces, another for infrastructure. None of them agreed. By the time the root cause was found, two hours had passed. Jonathan Trevor had seen this movie enough times to know that the plot was the problem, not the cast.
Trevor came to Observe from Wavefront, the cloud-native monitoring company that VMware would later acquire. Before that, he ran frontend engineering at Shocase, where he built the company's native iPhone app from scratch - a product problem, not just a code problem. The pattern shows: he's always been drawn to the experience of debugging, not just the mechanics of it.
"Observability is fundamentally a data problem - if you can bring all telemetry data together, curate it, and relate it, you can be an order of magnitude faster at detecting, investigating, and resolving issues."Jonathan Trevor, Co-Founder, Observe, Inc.
In 2017, Sutter Hill Ventures - the firm that also incubated Snowflake - brought together four engineers: Jacob Leverich from Splunk, Jonathan Trevor from Wavefront, Jon Watte from Roblox, and Philip Unterbrunner from Snowflake. They set up shop inside Sutter Hill's Palo Alto office and started building. Not a product, at first. A thesis. Observability was a data problem, and they were going to treat it like one.
The first product shipped in 2018, built on top of Snowflake's data warehouse. That choice wasn't accidental - it was structural. Observe believed that petabyte-scale telemetry data needed a petabyte-scale data foundation. Building on Snowflake wasn't a vendor dependency; it was an architectural conviction. When Snowflake announced its intent to acquire Observe in January 2026, the symmetry was almost too clean: the company that provided the foundation was buying the superstructure.
Observe emerged from stealth in October 2020 with a $35 million Series A. The pitch was simple and audacious: unified observability on an open data lake. Logs, metrics, and traces don't belong in separate systems. They belong together, cross-referenced, searchable, queryable - ready to answer the question your team is asking at 2am on a Tuesday.
By July 2025, Observe had raised $156 million in a Series C round, valuing the company at $750 million. The focus had sharpened around a new frontier: AI systems. LLMs fail in strange ways. Token costs spike without warning. Agent pipelines fork in unexpected directions. Observe built tooling to track all of it - token usage analytics, AI request failure analysis, LLM observability - before most of the market had even agreed on what "AI observability" meant.
Six months later, Snowflake closed the deal. Approximately $1 billion, its largest acquisition to date. Not bad for a company that started with four engineers, a shared office, and a disagreement with the status quo.