Product decisions, engineering deep-dives, and the thinking behind building an AI agent platform from scratch.
A technical deep-dive into Verslay's MCP server — how 156 tools are organized across 24 packages, how verslay_initialize + verslay_discover + verslay_deploy work, and the architectural decision to run no AI on our infrastructure.
How agency partners and individual resellers are building recurring revenue streams on Verslay's partner program — the economics, the tier structure, and what the Verslay Certified Reseller Badge actually means.
How Verslay's integration layer works — 13 OAuth providers, 7 always-on intelligence packages, token encryption with AES-256-GCM, and why centralized OAuth proxying changes the security model for AI-driven automation.
A step-by-step walkthrough of how Verslay's use-case system works — from picking a workflow to having 4+ agents running on your claude.ai session, with real integrations active.
Every business owner has Claude or ChatGPT open in a tab. Almost none of them are getting real work done with it. Here's why we built Verslay, and the specific bet we made on centralized memory, verifiable outputs, and deep integrations.
