Evidence-based conversational testing in Treseko
Design turn-by-turn conversational tests and review transcripts and expectations without confusing case format with execution mode.
This page is part of Treseko public resources for quality management, automation, evidence and release decisions.
Test the conversation as a journey
The case defines its goal and ordered turns, with expectations for evaluating each response. Treseko records the interaction so the team can review more than an isolated screenshot of the final message.
The first configured message is Turn 1. No historical greeting is inserted behind the scenes: message order is part of the test contract.
Format does not decide who evaluates
A conversational case can run manually, automatically, or with AI. Those are separate dimensions; mixing them leads to configuration mistakes that are hard to explain.
Persisted evidence may include the transcript, latency, HTTP details, assertions, and available traces. Eligible bugs are created from the saved execution, not from a conversation later replaced in the browser.
Explore each capability
View documentation · Explore QA solutions · Browse QA resources