PrendsTaPart incub IA

IA President is a multi-agent operating system that allows any founder to build, deploy, and operate AI agents using ready-made templates, MCP-connected workflows, and Gemini-driven configuration. Instead of building agents from scratch, users assemble them from curated templates mapped to MCP-driven automations, ensuring speed, reliability, and scalability.

Target users: solo-founders, incubators, agencies, SaaS builders, and small teams seeking AI automation without technical complexity.

Core value: Everything—workflows, storage, voice, memory, GitHub commits—runs through one unified Multi-Connector Protocol (MCP).
MCP triggers n8n workflows, pushes JSON configs to GitHub, interacts with Qdrant for memory, and handles voice via ElevenLabs.
The system becomes a Digital CEO managing a team of specialized manager agents.

---

⭐ Core Features

IA President (executive agent): interprets goals, selects templates, delegates to manager agents, validates outputs, and synthesizes.

Manager Agents (Tech, Product, Marketing, Ops, Finance): each adapts prompts, fields, workflow logic, UX structure, and KPIs.

Template Catalog: complete agent blueprints (roles, prompts, tools) mapped to workflow templates (lead management, onboarding, delivery ops, CRM sync…).

Gemini integration: reasoning, templating, content generation, workflow adaptation, and troubleshooting.

MCP Orchestration Layer:

triggers n8n automation scenarios (emails, CRM updates, notifications…),

pushes configs/workflows to GitHub,

reads/writes memory via Qdrant,

calls ElevenLabs for real-time AI voice,

manages external APIs (CRM, CMS, Drive…).

ReadyPlayerMe API: gives IA President a dynamic, 3D avatar interface.

---

🎬 Demo Flow (shown to judges)

1. User says:

“Create an AI Agent to manage my lead qualification and automate follow-ups.”

2. IA President selects the Lead Qualification Agent Template, connected to mapped n8n workflows via MCP.

3. Manager Agents ask a few targeted questions (CRM used, scoring rules, email tone).

4. Gemini adapts prompts, scoring logic, email sequences, and workflow parameters.

5. MCP activates the linked workflows:

n8n pipeline for lead intake,

automated email sequence,

scoring + CRM update,

logging to Drive or Notion.

6. Agent configuration, prompts, and workflow JSON are pushed to GitHub via MCP.

7. Qdrant stores long-term memory: tone, brand, CRM rules, previous outputs.

8. ElevenLabs + ReadyPlayerMe deliver a vocal, embodied demo:

“Your agent is ready. Shall we test it?”

9. User triggers a live test → all actions run through MCP.

10. IA President summarizes the agent, workflows, memory, and provides recommended next steps.

Previous Work
Before the hackathon, I developed several components that support this project:
• A digital incubator (PrendsTaPart) with a 13-step methodology used as structured knowledge for the agents.
• A suite of SaaS tools (RapidoCRM, RapidoCMS, RapidoRH) that provides real operational APIs the agents can call.
• A RAG memory system already used in my internal AI agents, which I am adapting for the multi-agent orchestration.
• A large knowledge base from my existing projects (Camicourse, EcoloCourse, Villa Beach, FoodEatUp, BraindCode), used to demonstrate real-world use cases.
All code for the hackathon is started from scratch today, but the conceptual foundations come from months of ecosystem building.