Product update: Tests—deploy with confidence, every time

Test every change against real runs. Ship knowing nothing broke.

Miguel Ribeiro

Until now, there was no way to know whether a workflow change broke something that used to work. 

You make a small tweak to a workflow, push it to production, and something doesn’t behave how you expected. A ticket that should have gone to Sales ends up in the wrong queue. A path that used to work now doesn't. And the worst part? You might not find out until a customer does.

That changes today with Tests.

For the teams building and maintaining workflows, Tests means you stop second-guessing every change and start shipping with a repeatable safety net. For ops leaders, Tests is proof that a change was validated before it went live—and a way to catch potential issues before they reach a customer, a queue, or a complaint.

See how Tests works


See it in action


Step through a live walkthrough of the product.



How it works

When you're developing a workflow on Noxus, you run it, see what it does, and iterate. When a run does exactly what you want, Tests lets you save it as a case—your benchmark. From that point on, every change you make is validated against those benchmarks automatically. Nothing breaks without you knowing.

Tests is built around three concepts that click together naturally.

A test suite is a collection of cases for a specific workflow or agent you want to validate.

Cases are real runs you've captured from that workflow or agent. Not synthetic examples constructed from scratch, but actual input-output pairs from real executions. This is your ground truth, built organically as you work.

Evaluators are the criteria you define to judge the output. Does the workflow route the case correctly? Does the agent respond with the right tone? Was the right action triggered at the end? Every evaluator runs against every case automatically.


The Tests interface—past runs that you can add as a case.


Let’s look at an example

Take a real example: a customer support workflow that receives incoming inquiries, identifies the client, categorises the case, and routes it to one of three paths—open a Sales ticket, let an agent reply with information, or hand off to a human for complaints.


The customer support workflow canvas in Noxus


As the team builds it, they run it against real scenarios. When a run routes correctly and produces a good reply, they save it as a case. Over time, they build a test suite that covers all three routing paths—Sales, Information, and Complaints.

Then they improve the categorisation logic. They update the flow, run Tests, and immediately see that two cases that previously routed to Complaints are now going to Information instead. Something that worked before now doesn't—caught before a single customer is affected.

Two complaints misrouted to Information—that would have meant two customers left waiting longer for a resolution they were already unhappy about. Tests caught it before it mattered.

They fix it, re-run, everything passes, and they ship.


The Tests results screen


That's the loop: iterate, capture good runs as cases, make changes, run Tests, know immediately if something broke, fix it, ship.

The same approach applies wherever you're automating complex decisions. For example, you can capture and validate:

  • A claims triage workflow that routes by complexity and urgency

  • An onboarding flow that generates personalised welcome communications

  • A dispute resolution process that escalates based on customer history

Anywhere a wrong output has a real consequence, Tests gives you a way to prove the right output still happens.

Why this matters

For builders, Tests collapses the feedback loop. Instead of running scenarios manually, eyeballing the output, and hoping nothing else broke, you run your suite, see what passed, fix what didn't, and ship. You get to production faster because you're not second-guessing—you're testing.

For operations and transformation leaders, every test suite is a living record of what your workflow is supposed to do. Every test run is evidence that it still does. When you're accountable for complaint handling, claims routing, or customer outcomes, "I think this works" isn't good enough. Tests gives you "I know this works, and here's the run that proves it."

Today, Tests is available for workflows. Support for agents is coming soon.

Whether you're building the workflow or are accountable for what it does—ship it knowing it works.

Miguel Ribeiro
Miguel Ribeiro
Cofounder & Senior Product Designer

Cofounder & Senior Product Designer

Connect with Our Team

You can also email us at sales@noxus.ai

Turn your customer Inbox into resolved processes

Trusted AI workers that gather evidence, apply policy, and execute audited actions — moving complaints, documents, and tickets from intake to done

Copyright ©2026, Noxus. All rights reserved.

Turn your customer Inbox into resolved processes

Trusted AI workers that gather evidence, apply policy, and execute audited actions — moving complaints, documents, and tickets from intake to done

Copyright ©2026, Noxus. All rights reserved.