Paved Marketplace MCP
From discovery to booking, all through Claude
Led end-to-end product development of an MCP integration that helped advertisers find the right newsletters for their audience and budget, and streamline sponsorship booking through conversation.
Problem / Context
Users struggled to find the best match for their audience and budget. They were doing a lot of things manually—comparing newsletters by hand, navigating fragmented metrics, and clicking through a lengthy, disconnected workflow across booking.
Goal & Success Metrics
Primary Metric
80% Accelerated Booking Time (streamlining discovery to booking)
Secondary Metric
Campaign Quality (sponsorship cost efficiency and performance)
Options, Decisions & Trade-offs
Initially considered building an in-platform chat assistant. However, considering the timeline, UI/UX complexity, legal considerations, and engineering effort, we decided to first launch via Model Context Protocol (MCP) in Claude. This allowed us to learn quickly from real user feedback and validate demand before deciding whether to invest in full on-platform chat infrastructure.
Learnings & Retrospective
What worked well
Going from ambiguity to a structured solution through close collaboration with engineering to deliver a high-quality product.
What I would do differently
Avoid implementing the full campaign journey upfront; start by launching and refining product discovery first.