
What happened
AWS made Amazon CloudWatch Omni generally available last week, with 17 built-in evaluators that score coherence, helpfulness, faithfulness and routing correctness, plus continuous scoring against live traffic.
Why it matters
Traditional monitoring can't tell whether an agent's answer was right, so the evaluation engine addresses the trust gap that keeps agents in pilot mode at enterprises.
What to watch
The IDE extension is free, but agents are chatty — every prompt and tool call generates a span — so ingestion costs can outpace the AI budget that created them. Watch the 30-day trial and $1,000 in OpenTelemetry ingestion credits.
WHO IT HITSEnterprise IT and AI platform teams running multiple agent pilots are the primary audience: they must now decide whether to standardize instrumentation on OpenTelemetry and model telemetry costs before agents go live, since Omni's evaluation and investigation features run on AWS.
Summaries like this, in your inbox every morning.
For years, observability tools answered a simple question: is it running? Agentic AI breaks that model. An agent can return a clean response and meet its latency target while still giving a customer the wrong answer or pulling from a stale knowledge base. AWS built Omni around that gap, moving observability outside the AWS Management Console, supporting OpenTelemetry and embedding AI so teams can start from a single trace and jump straight to evaluation.
The platform arrives in a crowded field. Datadog, Dynatrace, New Relic, Grafana Labs and Splunk are all expanding into agent observability, as are LangSmith and Arize AI. Omni's differentiator is its unified data layer: agent traces, application telemetry and infrastructure signals share the same CloudWatch data store, and AWS DevOps Agent is enabled by default in investigation sessions, correlating signals and keeping an investigation history that regulated industries may treat as audit evidence.
Openness helps, but lock-in isn't fully eliminated. OpenTelemetry makes data portable, yet the intelligence layer — topology, evaluators, investigation history and DevOps Agent — runs on AWS. Companies already invested heavily in AWS will likely default to Omni, while multicloud shops with mature Datadog or Splunk estates may use it for development and evaluation and integrate rather than replace. The outcome hinges on whether buyers treat evaluation as an operational discipline rather than a feature to switch on.
For example, today's edition would include:
AI-summarized, only the topics you pick: one digest a day via Email, LINE, or Slack.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Google said Gemini Omni 1.1 Flash is now directly available in Google Vids, letting users extend scenes while…

In a September 2025 paper titled "Why Language Models Hallucinate," OpenAI researchers said low-frequency fact…

The developer moved from IDE-centric work in WebStorm and PHPStorm to terminal-centric work with Claude Code…

In a preliminary code-review trial (internal log EXP-001-R2), Claude and Codex each independently found one va…

A developer moved Codex work to Pi Coding Agent, running gpt-6-sol at high thinking

While building an accounting app called Books tied to マネーフォワード, a developer built Jev Bookmarks
