How It Works
Connect store, choose plan, place snippet, go live.
The rollout path is intentionally compact: no heavy module-first dependency, no crawler load, and no sprawling frontend rebuild before the assistant is useful.
Rollout path
Move from setup to launch in a short, reviewable sequence.
Each step maps directly to a merchant control-plane surface and a launch-check item.
Connect Magento and sources
Add Magento credentials, import product context, and connect FAQs or policy content.
Apply recommended defaults
Provision widget identity, verification policy, localization defaults, handoff settings, and dashboard cards.
Install the storefront snippet
Place the widget snippet, verify live domains, and keep preview traffic separate from production heartbeat signals.
Enable protected support flows
Turn on verified customer sessions, personalization scopes, order support automation, and handoff policies as needed.
Enable proactive commerce scenarios
Turn on scenario types, configure per-scenario cooldowns, and choose a behaviour preset to reach customers at the right moment.
Launch and review analytics
Go live, watch verification conversion and support automation metrics, then refine prompts and policies without re-platforming.
Why the rollout stays light
The architecture is designed to keep operational cost predictable.
This is not a crawl-everything, sync-everything implementation.
No heavy extension required
Magento remains the source of truth while the assistant connects through APIs, selective sync, and storefront context.
On-demand reads
Protected customer and order lookups happen when needed instead of constant broad synchronization.
Compact safe caches
Customer and product caches store small summaries for runtime efficiency rather than raw oversized records.
Server-rendered admin
The control plane stays compact and fast without a large SPA dependency for every workflow.
Proactive commerce
Proactive scenarios activate as part of the same rollout.
No separate integration step. Once the widget is live, proactive commerce scenarios fire against session context, product signals, and cooldown rules already managed in the same control plane.
- Thirteen scenario types cover exit intent, re-engagement, low stock, personalisation, and more.
- Per-scenario cooldowns and a behaviour preset prevent over-messaging.
- CTA buttons inside the proactive bubble open chat in the right mode directly.