Playbooks · Coding tasks
Google's AQuA watches for silent agent failures
Written November 5, 2026
AQuA is an ambient quality agent that reads production traces to catch failures agents never report. What it checks, and what it can't see.
What AQuA is
AQuA, the Ambient Quality Agent, was released by Google on October 8, 2026 as an open reference implementation — composable building blocks you can run in your own project, not a managed service. It runs unattended beside your agent inside your Google Cloud project, sweeping production trajectories from Cloud Trace, Cloud Logging, or BigQuery on a schedule, after each deployment, or on demand. Session transcripts, source snapshots, and BigQuery tables stay inside your project boundary. AQuA never sits in the request path and never writes back to your agent.
The problem it addresses is the failure that leaves no trace of itself as an error. An agent can return HTTP 200, stay inside its latency budget, and raise no tool error — and still record a seat as confirmed without checking whether it is free, or recommend a steakhouse to a traveller whose profile says vegan. Health checks stay green; the failure is silent and happens inside the conversation.
What it checks
Each run reads a sample of up to 1,000 recent sessions and puts it through a five-stage pipeline:
- Sample. Pulls a random sample of up to 1,000 recent sessions from your telemetry, capped to keep run cost predictable.
- Review. Grades each session against a nine-point checklist of common breakdowns — procedure, tool selection and arguments, grounding, task completion, and others — with the agent's instructions, tools, and an optional developer goal file (
goal.md) in view. When a session fails, it writes a structured actual-versus-expected finding. - Cluster. Groups findings that share the same failure mechanism into candidate issue clusters.
- Verify. A separate model checks each candidate cluster against up to three full transcripts and discards clusters the evidence does not support.
- Track. Matches surviving clusters against open insights in BigQuery as new, recurring, or automatically resolved after 14 days unseen.
When an insight is worth investigating, root-cause analysis runs from the dashboard chat or from agents-cli aqua run. AQuA reads the failing trajectories alongside the immutable source snapshot captured at deploy time, cites the file and line range where the defect sits, and proposes an edit anchored to those lines. It never applies an edit or opens a pull request on its own.
Google demonstrates the pipeline on a travel-concierge agent with subagents for inspiration, places, flights, seats, and booking. Across a 32-session sweep, 5 sessions passed cleanly and 27 produced 42 findings, which clustered into 9 candidates; the verifier rejected 3 as false positives, leaving 6 verified issues. The largest was a seat-availability bypass in 15 sessions: the planning agent saved seat numbers straight into session state without calling the seat-selection subagent. A second was dietary constraints dropped across a subagent boundary in 7 sessions. After two one-line prompt fixes and a redeploy, replaying the same 32 sessions showed the seat bypass down 87 percent, from 15 sessions to 2, the vegan omissions down from 7 to 0, and full-session passes more than doubled, from 5 of 32 to 13 of 32. Those figures come from Google's own demonstration, not an independent test.
What it can't see
- Failures with no trace signal. AQuA reads trajectories. A failure that leaves no mark in the trace — or one that only shows up in a tiny slice of traffic — is hard to catch: the reference implementation samples up to 1,000 sessions per run at random, and Google names catching a silent regression that breaks 1 percent of a critical workflow as the harder problem.
- Long conversations, as-is. Passing whole transcripts works for ten-turn conversations and breaks down on coding or research agents that run for hundreds of turns and megabytes of tool output. Those need trajectory compaction first.
- External state. Turning a verified failure into a replayable test case works when tools are idempotent or mocked. Static replay does not reconstruct external state — a changed database row, a timed-out API, or a turn that depended on what the agent said two turns earlier will diverge.
- Your fleet, out of the box. The reference implementation is scaffolded one-to-one beside a single ADK agent in its repository. Attaching ambient quality sweeps across a fleet of agents from existing project telemetry is listed as future work, not something the release does today.
What to borrow without AQuA
You do not need AQuA running to take the habits behind it. Write down the failure modes you care about in plain language, the way a goal.md file does, and grade a sample of real sessions against that list instead of only checking that calls succeeded. Write each failure as an actual-versus-expected pair, so the fix is visible in the finding. And treat a cluster of failures as a claim until a second read of the full transcripts supports it — Google's design rejects unverified clusters before they enter the queue, and on an 87-trace internal benchmark its verifier rejected 4 of 24 candidate clusters.
To try the dashboard with no cloud project, no credentials, and no model calls, clone the adk-recipes repository and run make demo in the ambient-quality-agent directory, which serves the dashboard locally on a synthetic month of runs and insights.
Limits
- AQuA is a reference implementation released in the open, not a generally available managed product. Attaching it to your own agent means deploying its scaffolding in your Google Cloud project.
- The sweep numbers and cost figures in this page — a 96-session sweep at $0.70, the 32-session travel-concierge sweep at $3.76, root-cause diagnosis at $0.33 to $2.47 per insight — come from Google's own demonstration and its own pricing. They describe Google's example runs, not what a sweep of your agent will cost.
- No AQuA run is described here. This page follows the Google post and the repository it links to; it does not report a sweep of a production agent.
Related playbooks
- Google ADK 2.14: A2A graduates, Teams joins in
- Google Jules ends Nov 11: export your work
- 140 silent MCP contract changes, and how to pin them
- Antigravity's remote agents run in Google's cloud
Sources
- The Outer Loop, Insights First: An Ambient Quality Agent That Diagnoses Your Production Agent — Google Developers Blog
- adk-recipes — Google's agent recipes, including AQuA
Back to: Coding playbooks hub · Also: What AI monitoring missed, per the METR audit · Catch bad agent behaviour chains with Quarry