Airrived launches Agentic Observability control plane for AI agents
Airrived has unveiled Agentic Observability, a significant expansion of its enterprise Agentic OS, at the GISEC Dubai security conference. The Dublin, California-based company says the capability gives organisations end-to-end visibility into how autonomous AI agents behave, covering data ingestion, decision-making, and the business outcomes that agents produce.
The announcement lands against a backdrop of growing concern about autonomous agent governance. Anthropic has publicly urged the industry to accelerate AI safety measures to keep pace with agentic development, and Airrived is positioning this release as a direct response to what it frames as a critical governance gap: enterprises deploying agent networks that operate faster than any human operator can monitor.
What the platform does
At the centre of the product is a component Airrived calls Context Lake, a data fabric that aggregates context from enterprise systems and routes it through agentic applications. The Agentic Observability layer sits on top, surfacing four categories of information for each agent in a deployment: its creator and owner, the roles and permissions it holds, whether a human-in-the-loop approval step is required before an action executes, and what that action ultimately cost in token and model consumption.
On the risk side, the platform tracks the movement of personally identifiable information, payment-card data, and protected health information across agentic workflows. Airrived describes this as converting historically opaque sensitive-data exposure into a traceable, auditable chain. The economics component similarly converts token consumption into line-item cost tracking, which the company markets under the AI FinOps label.
Anurag Gurtu, co-founder and chief executive, said: "You cannot govern what you cannot see. Agentic Observability gives enterprises visibility from data, to decision, to action, to outcome."
Airrived did not disclose customer names, adoption figures, or pricing for the new capability in its release. The company supports on-premises, private-cloud, and air-gapped deployment models, which the release positions as a differentiator for regulated industries.
Market context and competitive landscape
Agentic observability is an emerging sub-category within the broader AI governance and MLOps market. Established players such as Datadog, Dynatrace, and New Relic have each announced or shipped tracing and monitoring extensions for LLM-based workloads, while specialist startups including Arize AI and Langfuse compete on model-observability and prompt-lineage tooling. Enterprise security vendors are also moving into the space: companies offering identity and access management for non-human identities are a natural adjacency, given that agents require scoped permissions much as service accounts do.
Airrived's framing around a unified "control plane" attempts to differentiate from point solutions by combining observability, governance, and cost tracking in a single layer. Whether that integration advantage proves durable will depend on how quickly larger platform vendors absorb similar functionality, a pattern that has repeatedly compressed the independent-tool category in adjacent markets such as container security and API management.
Regulatory read-across
The timing of the launch is notable given the EU AI Act's obligations around high-risk automated decision-making systems, which began phasing in this year. Agents that take consequential actions in cybersecurity, HR, or financial workflows are likely to fall within scope of those requirements, making audit trails and human-override mechanisms a compliance necessity rather than a product differentiator. The UK's AI Safety Institute has also signalled interest in agentic systems as a priority area for evaluation. Airrived's PII and PHI tracking features may carry additional relevance under GDPR and HIPAA where agents handle data subject to those frameworks.
The company's next visible milestone is the live showcase at GISEC Dubai. Investors and enterprise buyers will be watching for named customer deployments and independently validated benchmark data on agent-decision tracing latency and data-lineage completeness.