The monitoring system was having trouble staying alive. At Similarweb, engineers running an open-source logging stack faced at least one crash a month. Sometimes, new logs took roughly three hours to arrive. Consider the predicament: the evidence needed to explain a production incident was waiting inside another system that needed fixing.
- Logz.io runs the observability tools engineers would otherwise maintain themselves.
- Its commercial argument combines engineering time, useful telemetry and control over storage costs.
- OrionIQ extends the job to AI investigations, with runbooks and human approval controls.
Similarweb liked open source. It offered freedom to patch software and participate in a community. Its problem was the maintenance. In Logz.io’s published customer account, the costs of self-management and the managed service were comparable. The difference was what engineers could do with their day. After the move, the account reports monthly logging-stack incidents falling from at least one to zero.
The tool needed its own mechanic
That small absurdity explains a substantial software business. A company installs monitoring to understand its applications, then acquires another application to operate. Somebody must provision capacity, manage upgrades and recover from failures. Open-source code can be free while the responsibility attached to it remains expensive.
“We like open source.”
Or Tzabary · Similarweb
A preference that survived the decision to outsource its upkeep.
Tomer Levy and Asaf Yigal founded Logz.io in 2014. Levy later described their frustration with existing tools: too much complexity, too much cost, too much difficulty finding what mattered. The company began with managed ELK, the Elasticsearch, Logstash and Kibana combination used to collect and search machine-generated records. Its proposition was straightforward: keep the familiar tools; buy back the operating time.
Open source, with someone else on call
The scope expanded. Today, Open 360 brings together logs, metrics, traces and security data. Logs record events. Metrics track measurements over time. Traces follow requests through services. Logz.io’s foundations include OpenSearch, Prometheus and Jaeger, with OpenTelemetry providing a standard way to instrument applications and move telemetry.
The customers include SREs, DevOps engineers and security teams. Public examples stretch from ZipRecruiter and Unity to Siemens and Snyk. Logz.io sits between maintaining an internal stack and buying managed observability from vendors such as Datadog, New Relic or Grafana Cloud. Its expertise combines operating the underlying systems with investigation workflows, integrations and telemetry economics.
There is a useful complication in its open-source identity. In July 2024, Logz.io replaced its OpenSearch Dashboards-based log interface with Explore. Yigal said customer feedback drove the change. Familiar foundations were valuable, but the inherited interface could get in the way. Commercial SaaS built on open source still makes its own product decisions.
The company also reaches customers through their cloud providers: AWS Marketplace availability and a Microsoft partnership announced in 2021 let cloud customers provision observability through existing purchasing channels. Its own careers page emphasizes urgency, alignment and ownership. Those are employer promises; the product’s more tangible expression of ownership is taking operational chores off a customer’s list.
Investors financed that widening brief. General Catalyst led a $52 million Series D in 2019. A $23 million round led by Pitango Growth followed in November 2020, bringing disclosed total funding above $120 million. The smaller follow-up was deliberate: Levy told CTech that much of the previous round remained available, while pandemic uncertainty made additional capital attractive.
The invoice has a memory
Logz.io earns money through subscriptions and consumption plans. The bill reflects telemetry volume and retention, with separate AI usage charges. Published US East log pricing starts at $0.92 per ingested gigabyte with seven days’ retention under annual billing. Monthly billing costs 1.2 times the annual rate. Region, negotiated volume and storage choices change the total.
1 GB ingested daily × $0.92 × 30 days.
Illustration using the published annual-billing log rate and seven-day retention. Excludes AI, extensions and other products.AI has its own meter. Documentation lists AI Chat at $10 per million tokens and Alert Analysis at $10 per invocation; other agents vary. Spending caps are available. Across telemetry, filtering and storage tiers let teams decide which information deserves expensive, readily searchable retention.
That choice has consequences. Consumption-plan documentation says reaching an ingestion cap pauses collection until the limit changes or the day ends. A predictable bill can therefore produce a gap in the evidence. Teams should model a noisy incident, longer retention and AI investigations together before treating a headline unit price as their budget.
Teach the agent where to look
In April 2026, Logz.io launched OrionIQ after six months in beta with dozens of customers. Agents can investigate alerts and follow operational workflows, using telemetry and organizational context. Teams determine which actions run autonomously and which require approval. A May expansion added coordinating agents, incident triage and connections to tools including Jira, GitHub, Slack and Confluence.
ThetaRay’s customer account supplies a revealing detail. The financial-crime technology company centralized logs and metrics before piloting OrionIQ. In one investigation, an engineer initially dismissed a cause as unrelated. After a runbook change told the agent to investigate deeper, it identified the correct cause. The instruction mattered. ThetaRay’s larger ambition of a largely autonomous operations center remains a goal.
The lesson readers can copy is modest: choose a recurring investigation, collect its evidence, write its decision points down, then measure the result. This approach needs usable instrumentation and current instructions. Missing telemetry or a stale runbook leaves the investigator with the same handicap as its human colleague. Logz.io’s bet is that those instructions, joined to the right evidence, can give engineers fewer reasons to open a laptop at night.
There is a second AI job here, too: monitoring the agents customers build themselves. Logz.io’s early-access AI Observability product traces model requests and tool calls alongside backend services. It uses OpenTelemetry and OpenLLMetry rather than requiring a proprietary SDK. The distinction matters to buyers: automating an investigation and examining a misbehaving AI application are separate tasks, even when both live beside the same telemetry.