An embeddable AI assistant that qualifies customer enquiries in conversation and hands the team a structured brief.
Business challenge
Service businesses lose enquiries when visitors face long forms or wait for a reply, and staff receive unstructured messages that need follow-up questions before anyone can quote.
Solution
A conversational front office that qualifies the request in the visitor's own words, confirms the captured details with the visitor and delivers a structured brief, with the business-specific logic isolated in vertical packs and company configuration.
Key capabilities
Product showcase
Conversation visuals replay recorded model replies from a local test build, using generic demo configurations only.
How it works
In the chat widget, in their own words.
It collects the relevant details step by step and captures them into structured fields.
A review card shows what was captured; the visitor confirms or changes it before sending.
A structured enquiry with a suggested next action reaches the team.
Technical approach
A TypeScript Node.js service with an Express API and SQLite storage, using Google Gemini through the official SDK. A mode switch separates production, live-test and free stub modes, and the stub mode refuses any non-local endpoint. Browser tests run the real widget against a mock model server; recorded model responses are replayed as fixtures so behaviour can be regression-tested without model cost.
Status and limits
Related solutions
More selected work
A multilingual corporate website for international business infrastructure, generated by a validated content engine and paired with an AI pre-check assistant.
A multi-module operations platform that gives one owner a single focus queue, decisions, tasks, portfolio and lead inbox across several business directions.
A custom platform that turns manual B2B prospecting into a structured, auditable workflow: research, qualification, outreach drafts prepared for human review, and clear status tracking.
Tell us what is not working today. You get a straight answer on what it would take, and whether we are the right team to do it.