MVP

AI / Business Automation / Agentic AI

AI Business OS is a hackathon-MVP command center where a CEO issues a plain-language directive, DAPA (the orchestrator) classifies and delegates it to AI employees — a Sales AI and a Receptionist AI — and any external or scheduling action is gated behind explicit CEO approval before it happens, with every decision and outcome recorded in an immutable audit log.

Running the day-to-day of a small business means coordinating multiple disconnected tools and staff — and handing that coordination to AI agents only works if a human stays firmly in control of anything consequential.

A Fastify backend (the Business OS Core API) runs the DAPA orchestrator and a provider-agnostic Model Router spanning Anthropic, OpenAI, Gemini, DeepSeek, and local Ollama, routing directives to AI employee adapters behind an approval gate and an immutable audit log; a Next.js 15 CEO Dashboard provides the command-center UI, including live executive-summary KPIs/charts and a chat interface grounded in live operational data.

DAPA orchestrator classifies and delegates plain-language CEO directives to AI employees

Explicit CEO approval gate before any external or scheduling action is taken

Provider-agnostic Model Router across 5 LLM providers, including local Ollama

Immutable, append-only audit log of every decision and outcome

CEO Dashboard with executive-summary KPIs/charts and live-grounded chat

Spec-driven development — foundation specs (vision, PRD, architecture, roadmap) and a formal hackathon-MVP feature spec preceded implementation.

A hackathon MVP built and run locally alongside its sibling AI-employee projects (ai_sales_fte, receptionist_fte) — not yet deployed publicly, so no live URL is claimed.

Outcome data not yet measured for this project.

Founder & AI Systems Designer — architecture, backend, and frontend

FastifyNext.jsTypeScriptPostgreSQLRedisAnthropic / OpenAI / Gemini / DeepSeek / Ollama