Scalable Chat Commerce SaaS Architecture
Architecting a scalable conversational commerce engine using Nuxt 3 and Express js to turn standard WhatsApp chats into a fully functional sales channel.
The Challenge
Serving 1,000+ business users, the DiChat.net SaaS platform (built on Laravel and Node.js) struggled with crawl efficiency and poor indexation. Beyond the technical architecture not being optimized for search engines, the product itself needed to solve a harder problem: letting businesses run full commerce operations, from customer conversation to checkout, without ever asking the customer to leave WhatsApp or install a separate app.
The Solution: Scalable Technical Architecture and In-Chat Commerce
To unlock organic growth and support the product’s core value proposition, I led a comprehensive technical SEO overhaul alongside the platform’s AI and commerce architecture.
1. Crawl Efficiency and Indexation
I conducted a deep technical audit to resolve server-side rendering bottlenecks and JavaScript execution issues. By restructuring the internal linking architecture and optimizing server response times, we drastically improved how efficiently Googlebot could crawl and index the platform’s core landing pages.
2. Landing Page Optimization
We rebuilt the primary landing pages to align with specific commercial intents, using semantic HTML structuring, optimized meta tags, and content that directly answered the pain points of B2B users looking for WhatsApp API commerce solutions.
3. AI-Driven Conversational Commerce
Beyond discoverability, I designed the system architecture that lets DiChat function as a replacement for human customer service agents in routine conversations. This included:
-
A switching layer that routes conversations between rule-based automation (for predictable, high-volume flows like order status or FAQs) and a custom LLM-driven agent for conversations that need context, judgment, or multi-turn reasoning, so businesses aren’t forced to choose between cheap automation and capable automation.
-
A custom LLM implementation with persistent conversation memory, built to handle customer inquiries with the continuity a human agent would have, rather than treating every message as a stateless query.
-
End-to-end structured commerce inside a single WhatsApp thread, product browsing, cart, and checkout, without the customer ever needing to install a separate app or leave the conversation they’re already in.
The Impact
The technical overhaul and product architecture together transformed the platform into a scalable organic growth engine with a genuinely differentiated product experience.
-
Improved Indexation: Core product pages achieved rapid indexation and higher SERP placements.
-
Scalable Growth: Built a robust technical foundation that seamlessly supports the platform’s 1,000+ business users while continuously capturing new organic leads.
-
Reduced Dependency on Human Agents: The switch-bot and LLM architecture allows businesses to automate routine and complex customer conversations alike, without the friction of directing customers to a separate app.