AI & predictions
Predictive Immersive Analytics
Predictive immersive analytics applies machine-learning models to spatial telemetry to forecast drop-off, comfort issues and completion likelihood before they occur at scale.
Instead of post-hoc reporting, predictive immersive analytics surfaces the scenes most likely to break comfort or retention in your next release. Gossip Analytics uses this layer to prioritise fixes by predicted impact.
The underlying models are trained on aggregated cross-project data plus the studio's own historical releases. The output is not a black-box score but a ranked list of scenes with the specific signal patterns driving each prediction, so designers know what to act on rather than what to trust.
Why it matters
By the time a comfort or retention problem shows up in traditional post-hoc analytics, the release is already public and the damage is already done. Predictive immersive analytics is what moves those problems from post-mortem to pre-release, which is where studios can still act on them.
Example
Before a soft-launch, the predictive model flags scene 4 as the highest drop-off risk based on a combination of comfort trend and hesitation clustering. The team redesigns scene 4's transition; the actual soft-launch shows drop-off in line with best-case scenes, not with the predicted risk.