On-site chat
A branded widget answers sizing, stock and shipping questions using your live catalog, then closes the sale instead of collecting an email address.
Guide
Conversational AI for ecommerce
Online stores lose revenue in the gap between a shopper's question and a reply. Conversational AI closes that gap by answering in the channel the shopper already chose — chat, text or phone — and by doing the work behind the answer, not just describing it.
One brain, every channel
Phone calls
An always-on voice concierge greets callers by name, reads back their order status, and hands hot callers to a supervisor with the case already summarised.
SMS and email
Two-way texting and inbound email run through the same brain, so a shopper can start in chat and finish by text without repeating themselves.
What good conversational AI actually does
Resolves, not deflects
Order lookups, address changes, return labels and refunds are executed against your store, inside the caps you set.
- Live order and tracking lookups
- Prepaid return labels
- Refunds released on first carrier scan
Grounded in your data
Answers come from your catalog, policies and past resolved cases through vector retrieval with citations — not from a model's guesswork.
- Product and policy embeddings
- Cited sources on every answer
- Refuses to guess below your confidence floor
Measured like a contact centre
Every conversation is scored on empathy and policy adherence, timed against your SLA, and rated by the shopper.
- Resolution rate and CSAT
- SLA breach alerts
- QA scorecards and coaching
Works with your platform
Shopify, WooCommerce, Square, BigCommerce or a custom Stripe setup — one adapter, the same conversational layer on top.
- Five commerce platforms
- Secure payment links, no card data spoken
- Sandbox mode before you connect keys
How to roll it out
- 1
Connect your store
Katalyst reads your catalog, policies and order history so answers are tied to real inventory.
- 2
Set the guardrails
Choose refund ceilings, a confidence floor, and whether the concierge drafts for approval or acts on its own.
- 3
Open one channel
Start with on-site chat, review the transcripts and QA scores for a week, then add SMS and the phone line.
- 4
Escalate with context
When a case outgrows the concierge, your team receives a dossier: what happened, customer value, mood and a recommended action.
Frequently asked questions
What is conversational AI for ecommerce?
It is software that holds a natural back-and-forth with shoppers across chat, SMS and voice, and can act on their account — checking an order, starting a return, or booking a call — rather than only answering questions with canned text.
How is it different from a chatbot?
A chatbot follows a decision tree and hands off when it runs out of branches. Katalyst reads the shopper's intent and sentiment, retrieves your real catalog and policy data, executes the action, and escalates with a written brief only when the case needs a human.
Will it invent answers about my products?
No. Answers are retrieved from your own catalog, policies and resolved cases with citations. Below the confidence floor you set, it defers to a person and logs the knowledge gap for you to answer once.
How long does setup take?
Create a workspace, connect your store, and the catalog and policy sync runs on its own. Most stores are answering live chat the same day, and the Setup Academy walks through the phone line and escalation rules step by step.
Put conversational AI on your storefront
Create a workspace, connect your store and watch the concierge handle its first order question today.