Products · Exploring
Conversation evaluation
More and more organizations put conversational agents in front of their customers, but few know how to measure whether they talk well. We're exploring a platform that simulates conversations, detects failures and helps teams review and improve continuously, with special attention to Spanish and its accents.
Find where your assistant gets lost, before your users do.
What we're exploring
Simulate
Test conversations with different user profiles, accents and difficult situations.
Detect
Misunderstandings, made-up answers, loops and handovers that never happen.
Review
Expert human review to decide what to fix first.
Where it stands
Where we are
The platform grows out of the evaluations we do as a service, such as the two chatbots of a real estate portal or the review of Clevergy's voice assistant. Today we evaluate with our own tools and expert human review; we're exploring how to turn that method into a platform any team can use continuously.
Who it's for
Product teams with an assistant or agent live or about to launch; regulated sectors or services with vulnerable users, where a failure has consequences; teams working in Spanish who need to test its accents and varieties.
What it doesn't do
It doesn't replace human review or certify compliance with the EU AI Act: it helps you find problems and decide what to fix first.
How to pilot it
We start with an evaluation of your assistant as a service and test the platform on your conversations. You keep the diagnosis; we learn what the tool needs.
Related services
In the lab
Running an assistant or an agent? We're looking for organizations to pilot the platform with.