Feature groups
Every feature ties back to service quality, conversion, or operating leverage.
This page is structured around business problems first: repetitive support work, product discovery friction, post-purchase confusion, workflow overhead, and conversion opportunities beyond the widget.
Support automation
Recommendation optimization
Workflow automation
Proactive commerce
Feature map
Organized around the full customer journey.
The site should make it obvious that the agent does more than answer generic support questions.
AI customer support
FAQ answers, policy guidance, and how-to replies in a consistent brand tone.
Recommendation optimization
Trending, affinity, boosts, exclusions, slot controls, and recommendation analytics.
Order tracking
Order-status, delivery estimates, and tracking communication without a separate support ticket.
Returns automation
Policy clarity, return steps, and refund expectation guidance.
Workflow automation
Trigger → rule → action flows for routing, escalation, follow-up, and operational alerts.
Proactive commerce
Twenty scenario types with fast-path FAQ, reassurance, and session-aware nudges.
Growth surfaces
Shareable advisor pages, AI wishlists, and a public assistant route beyond the widget.
Multilingual communication
Reply in the customer language while keeping one operational setup.
Human handoff
Escalate low-confidence or high-risk cases with full context.
Analytics
Track support load, resolution patterns, and operational blind spots.
Integrations
Magento, feeds, APIs, storefront context, and workflow hooks.
Customization
Tone, policies, escalation logic, and business rules shaped around the brand.
AI customer support
Problem: the queue fills with repeat questions. Feature: an always-on agent with approved answers. Benefit: fewer tickets and faster service.
This is the core support automation layer, built for FAQs, policy answers, and common pre-purchase or post-purchase questions.
- Answer shipping, sizing, payment, warranty, and return-policy questions from approved content.
- Keep replies clear, on-brand, and easy to audit.
- Escalate when the question needs human review or higher-confidence handling.
Live Chat
Online
Recommendation optimization
Problem: support conversations often end without a buying nudge. Feature: recommendation controls and analytics. Benefit: more revenue from existing traffic.
Recommendations should feel relevant and useful, not random. That means connecting the assistant to real product data, ranking controls, and measurable outcomes.
- Use product titles, descriptions, attributes, categories, FAQs, and availability data.
- Support alternatives, bundles, and accessory suggestions.
- Apply trending signals, affinity, boosts, exclusions, and slot caps without code changes.
- Measure impressions, clicks, CTR, and conversions per recommendation strategy.
Recommendation Engine
Active
Trending last 24 h
1
Trail Jacket L
★ 98
2
Daypack 20L
★ 91
3
Trek Gloves
★ 84
Order tracking and returns
Problem: post-purchase questions generate avoidable service load. Feature: guided self-service for tracking and returns. Benefit: lower support cost and a better customer experience.
The assistant stays valuable after checkout by helping customers understand where the order is and how returns work.
- Resolve common “where is my order?” questions quickly.
- Set delivery expectations clearly and calmly.
- Guide return requests with transparent policy and refund timing language.
Order #MG-48221
On the way 📦
Order confirmed
Mar 11 · 09:14
Shipped with DPD
Mar 12 · 14:30
Out for delivery
Today · In progress
Delivered
Expected today by 18:00
🚚 Estimated delivery: today before 18:00
Workflow automation
Problem: support teams still lose time on routing and follow-up. Feature: one automation layer across the workspace. Benefit: fewer manual touches and clearer operational rules.
Workflows keep escalations, notifications, and queue routing visible as explicit rules instead of buried conditional behavior.
- Use starter templates for common escalation and lead-capture flows.
- Route by trigger family, add tags, assign queues, and notify internal owners.
- Keep automation explainable to support leads, operators, and reviewers.
Workflow automation
Live
Trigger
return_requested
Conversation event received
→
Rule
if confidence is low OR return policy applies
Run the explicit automation policy
→
Route to Returns queue
Tag conversation: return_flow
Notify support lead
Proactive commerce
Problem: shoppers leave before they ask for help. Feature: twenty session-aware proactive scenarios. Benefit: earlier intervention without pop-up spam.
The proactive layer scores page context, dwell, funnel stage, and product signals to decide when to show inline FAQ, reassurance, comparison help, cart nudges, or exit protection.
- Twenty scenarios across V1 and V2, including decision assist, cart insight, and soft exit catch.
- Fast-path logic on PDP can show inline FAQ first and confidence booster as fallback.
- Per-scenario toggles, presets, cooldowns, and analytics keep the layer under merchant control.
Page signal
product · 32s dwell
Viewed
3 products this session
Best choice scenario
●
💬
Based on what you've been viewing, the Trail Jacket L is the top pick — want me to explain why?
Growth surfaces
Problem: good recommendation sessions disappear when the chat ends. Feature: public pages generated from live shopping context. Benefit: more shareability, more entry points, and more conversion reuse.
Growth surfaces extend discovery beyond the widget with public advisor pages, wishlists, and a public assistant entry point that still uses the same catalog and recommendation logic.
- Shareable product advisor pages for side-by-side decision support.
- AI wishlist pages for gifting, save-for-later, and collaborative buying flows.
- A public store assistant page that captures discovery traffic outside the storefront widget.
Growth surfaces
Publishing
Advisor page
3-product comparison page
124 views
AI wishlist
Shared for gifting and save-for-later
39 product opens
Public assistant
Public route with live catalog context
11 add-to-cart actions
Trust and control
The product stays explainable to the people who actually carry the risk.
That includes support leaders, ecommerce managers, finance teams, and legal reviewers.
Human handoff
The team stays in control when confidence is low or the request is sensitive.
Auditable workflows
Operational automation stays visible as explicit rules instead of hidden branching.
Billing clarity
Trials, renewals, cancellations, and charges stay easy to understand before checkout.
Analytics and oversight
Track what the assistant handles well and where operators still need to step in.