← Back to RicardoOS
Building2026

AI Strategy Table

A cinematic multi-agent decision room.

Next.jsTypeScriptVercel AI SDK

AI Strategy Table is what happens when you stop asking one model for one answer and instead sit four of them around a table. You type a strategic question, and four AI advisors with different temperaments deliberate on it in a Brass & Neon war-room UI. A moderator then reads the room and hands you a single decision brief instead of four competing walls of text.

The interesting constraint was money. It would be easy to let agents chatter forever and watch the bill climb, so the whole thing is stateless and deliberately cheap: a full session is two or three model calls, no more. Every response is structured output validated with Zod, which means the UI never has to guess what a model meant, and I never have to parse prose by hand.

It runs on Next.js and TypeScript, with the Vercel AI SDK orchestrating the calls, Zustand holding the session together on the client, and Tailwind carrying the war-room look. It's less a chatbot and more a decision instrument that happens to think out loud.

Lamadrid Labs © 2026