Glance Intelligent Detect
← Decision pack
Product journey · MVP

How Glance Intelligent Detect works for customers

Not a playbook CMS and not “describe intent → AI publishes an offer.” Events are necessary but not sufficient; golden paths define what gets evaluated, then the coach guides users in-product.

Interactive console: gid-admin-mock.html · Landing: gid-landing.html · Docs: Quickstart

Is it automatic if we only send events?

No. Glance Intelligent Detect still needs you to say which workflows matter, then it labels sessions and (with triggers) decides when to speak.

If you only…You get…You still need…
Stream events Live activity pipe Workflow ontology (golden paths)
Define golden paths Sessions scored vs ideal sequence Triggers + delivery (coach / chat / tour)
Connect a support agent Context-aware reactive + proactive chat FSM / cooldown so it doesn’t spam

Why the console exists

Admin is not primarily “write CSS steps” or “prompt an offer into existence.” Deep trigger authoring can live in SDK / eng config.

Step-by-step (MVP)

Click through the mock in this order.

1

Connect activity

Open the activity pipe before anything “smart” happens.

MVPPaste SDK snippet → wait for / simulate first events → feed live. Analytics webhooks come later.
2

Define golden paths

Tell the system which journeys matter. Glance Intelligent Detect labels sessions against them.

MVPRecord path → perform the happy path in your app → save. Toggle Evaluating / Paused. Same event stream — no second collector.
3

Live, evaluate, coach

Sessions show workflow + completion %. Off-path / stalled can fire proactive help.

MVPLive feed of labeled sessions + proactive chips. Coach in the product renders guidance JSON (spotlight + sheet). Agent MCP and tour providers: Coming soon.

Out of primary path (for now)

Those may exist later as advanced / eng escape hatches — not the default happy path.

Open console mock Landing Docs Quickstart GID × Prototype