What makes Autoplay session analysis different?
Autoplay session analysis is built around what agents and copilots need to personalise to each user. It's designed to account for three core user behaviours:
- Every user is their own edge case — most tools assume everyone follows the same path; Autoplay doesn't
- Users don't know what they don't know
- Users don't want to be told what they already know
After extensive research, we found that the most important signal to extract was which workflows a user had completed — and at what completion rate. This became the foundation of our approach: every session replay is labelled by workflow.
How the labelling works
- You define all the tasks and workflows in your product
- Autoplay runs your session replays through TERRA
- Each session is output with a workflow label and completion rate — see the full schema at developers.autoplay.ai/sdk/user-memory
TERRA takes each workflow and its completion rate as inputs. It uses a combination of computer vision and raw JSON (rrweb) to normalise user actions, then compares them against the tasks and actions you've defined.
