AI Assistant Platform

Jun 2025 – Apr 2026

AI Assistant Platform

Co-developed with three engineers to embed custom AI assistants into organisational websites. RAG with vector search via Qdrant, dynamic tool calling, and a seamless drop-in widget.

View Live
Next.jsQdrantPostgreSQLRedisNode.js

Screenshots

Inside the product

AI assistant chat widget

AI assistant chat widget

Embeddable chat widget that drops into any site with a single script tag. Supports streaming responses, source citations, and brand-matched styling out of the box.

Knowledge base configuration

Knowledge base configuration

Admin panel for uploading documents, chunking content, and syncing embeddings into Qdrant. Teams manage knowledge sources without touching infrastructure.

Vector search and tooling panel

Vector search and tooling panel

Tool-calling configuration where assistants invoke APIs, query databases, and chain actions. Vector search settings tune retrieval quality per use case.

Assistant analytics dashboard

Assistant analytics dashboard

Usage analytics covering conversation volume, resolution rates, and tool invocation patterns — helping teams iterate on prompts and knowledge coverage.

Project Highlights

  • Architected RAG pipelines with Qdrant vector search for context-aware assistant responses.
  • Built a drop-in widget that organisations embed with minimal integration effort.
  • Implemented dynamic tool calling so assistants can query live data and trigger actions.
  • Collaborated across the full product lifecycle — from system design to production deployment.