QA Hub: Agent quality is a team sport

In the era of AI agents, CX teams gain a continuous stream of real-world signals about what customers need and where the experience can improve. Automated QA helps surface patterns across that growing volume, while frontline teams help define what good looks like in new or nuanced scenarios. Human feedback is especially valuable for interpreting edge cases, pressure-testing new rubrics, and bringing distinct perspectives into the review process.

QA Hub gives CX teams a more streamlined way to manage human-in-the-loop quality review, with a dedicated workspace to organize QA batches, apply custom rubrics, and track follow-ups. The teams responsible for evaluating and improving AI agents deserve purpose-built tools, and QA Hub gives them a best-in-class workflow for turning human feedback into agent improvements.

Human feedback deserves a first-class workflow

No two CX teams run QA the same way. Some organize reviews by assignee, others by queue, issue type, severity, or custom rubric fields. QA Hub gives teams the flexibility to shape the workspace around their own process, with configurable columns, filters, and saved views that make it easy to return to the conversations that matter most.

QA also moves faster when the right people can work together in one place. Teams can organize batch reviews, assign owners, tag teammates directly in conversations, and stay aligned on what needs attention. Notifications help make sure comments, assignments, and follow-ups don’t get lost.

When QA needs to include people outside the Decagon platform, the External QA Page makes that possible too. Teams can share standalone QA pages for testing or QA blitzes, complete with profiles, mock tool responses, chat and voice testing, live voice transcripts, and session history for reviewing previous QA runs.

Manual and automated QA work better together

QA Hub and Watchtower are complementary tools that help teams answer the same core question at different levels of scale: where is the agent performing well, and where does it need to improve?

QA Hub is most valuable when teams need human expertise to build trust, pressure-test new rubrics, or bring multiple perspectives into the review process. Watchtower gives teams broad coverage once they know what to monitor, applying proven rubrics across conversations, scoring interactions consistently, and surfacing recurring risks and regressions at scale.

‍

Together, QA Hub and Watchtower give teams both the depth to understand quality through human feedback and the breadth to monitor it consistently at scale.

Duet helps teams turn feedback into action

Duet helps CX teams move faster across the QA process, from creating review batches to synthesizing what reviewers find. Teams can ask Duet to sample conversations and create a QA batch, such as 50 random conversations from the past week or a set of conversations assigned by intent or insight tag. Reviewers can then triage and score those conversations directly in QA Hub. Duet can also take on the review itself, scoring conversations and surfacing results in QA Hub to proactively identify issues and give teams continuous coverage without manual effort.

Once feedback has been captured, teams can ask Duet questions like, “What are the most common issues flagged during QA?” Duet analyzes reviewer feedback, clusters recurring issues, and generates reports with common issue categories, severity breakdowns, and supporting evidence. This helps teams move from manual review to a more systematic improvement process, with a clearer view of which AOPs, guardrails, guidelines, or knowledge sources to improve next.

A better operating rhythm for agent quality

The best agent quality programs are collaborative, combining the scale of AI with the expertise of the people who understand customers best. QA Hub gives teams a first-class workflow for that collaboration, helping them review conversations, surface patterns, and turn feedback into better agent performance.

To see QA Hub in action, request a demo with our team.