01
Aug 2025 - Present
C-Link
Multi-agent AI document automation for construction SaaS
Engineering Team Lead · Xynotech
Problem
Client onboarding for a construction SaaS platform depended on manually turning contractor and quantity-surveyor Word documents into dynamic templates, slowing delivery for clients including McLaren.
Approach
Led 5-8 engineers from SPIKE and architecture through sprint execution, reviews, and releases. As a proof of concept directly assigned by the CTO, independently architected a multi-agent AI pipeline: uploaded DOCX files are converted to HTML and rendered in a WYSIWYG editor, one AI agent scans the document to identify and de-duplicate candidate dynamic fields (shortcodes), and - once a field is approved or a structured change request is submitted - a second AI agent drafts a developer-ready requirements brief with a suggested implementation strategy, logging every step to Jira via API.
Outcome
40% less time spent on manual document processing, with finished documents published as live dynamic templates for clients including McLaren.











