ChatWith.store
AI-powered conversational commerce platform
An AI sales assistant for e-commerce stores that combines conversational product discovery, live inventory intelligence, product recommendations, customer support, cart building, and checkout assistance.
01 / CONTEXT
The problem
AI commerce systems must reason over frequently changing inventory, pricing, availability, product variations, delivery terms, and order state without introducing stale or incorrect information.
02 / APPROACH
How I built it
The AI layer handles natural-language understanding and recommendation, while live commerce operations are delegated to structured tools connected to Shopify and WooCommerce APIs.
03 / CONSTRAINTS
What had to hold
Live inventory Dynamic pricing Product variants Multiple commerce platforms Low-latency conversational interaction Accurate transactional information
04 / OBJECTIVES
Definition of done
• Improve conversational product discovery • Provide live product information • Recommend relevant products • Reduce shopping friction • Enable conversational cart building • Connect shoppers directly to checkout
05 / THE SYSTEM
Architecture
Headless conversational interface A chat interface can be embedded into store pages or connected to messaging channels. Tool-enabled agent The AI uses structured tool calls for commerce operations. Live inventory access Price and availability are fetched directly from merchant systems. Streaming communication WebSocket/SSE communication supports responsive conversational experiences.
06 / DECISIONS
Key decisions
Do not embed volatile inventory in model context Why: Product availability and prices can change too quickly for static model context to remain trustworthy. Use deterministic commerce tools Why: The AI should query authoritative store APIs for transactional information. Make product actions conversational Why: Product discovery should lead naturally into cart and checkout actions.
07 / SHIPPED
Deliverables
• Conversational sales assistant (in-progress) — Core capability of the product (see features). • Semantic product search (in-progress) — Core capability of the product (see features). • Live inventory synchronization (in-progress) — Core capability of the product (see features). • Product recommendations (in-progress) — Core capability of the product (see features). • Cross-selling (in-progress) — Core capability of the product (see features).
08 / LESSONS
What I'd keep
• Sales AI should guide shoppers toward actions rather than merely answer questions. • Live commerce data should come from deterministic tools rather than static model context. • Product chips, clear calls-to-action, and low-friction cart actions are critical to conversational commerce.
What changed
Higher
Engaged conversion rate
to date
Instant
Shopper response
to date
// START A PROJECT
Building something with similar constraints?
If it's in your critical path and you'd rather not learn its failure modes live, let's talk about it.