The Complete Picture
Infrastructure costs, pricing architecture, unit economics, profitability scenarios, 18-month milestones, and cap table — every number behind the investment thesis.
The Largest Underserved Market in AI
The professionals who need AI most are legally forbidden from using it.
Healthcare
Patient records cannot leave hospital infrastructure. HIPAA and GDPR make cloud AI illegal for clinical use — fines reach €20M per breach.
Legal
Attorney-client privilege prohibits uploading case files to any external service. A single breach ends a career — or a firm.
Finance
Portfolio strategies, client data, and M&A intelligence are MiFID II-regulated. Regulators do not care how powerful the AI is.
"I'd love to use AI for diagnosis support, but I can't upload patient records to any external service. It's illegal."
— Dr. Elena M., Psychiatrist, MilanPRISM: Your Professional AI Twin
A self-hosted platform creating AI agents trained on your knowledge, writing in your voice, running entirely on your infrastructure. Six distinct layers — each a deliberate architectural choice.
Identity Layer
Built through a guided wizard. Defines domain, role, tone, language, citation rules. No prompt engineering required.
Knowledge Layer
Documents become expertise — protocols, regulations, case files, research. Anti-hallucination by design: the Twin knows what it doesn't know.
Style Layer
PRISM extracts your voice from your own documents — vocabulary, structure, formality, signature phrases. Outputs read as if you wrote them.
Reasoning Layer
Cross-domain synthesis, pattern recognition, conflict detection, full source attribution. Not retrieval — reasoning across your entire knowledge base.
Memory Layer
Persistent dual memory: short-term session context and long-term conversation history. Nothing is ever lost between sessions.
Learning Layer
Conversations generate validated insights that accumulate as permanent professional knowledge. The platform grows more precise with every interaction.
Built by the Person Who Understands the Problem
The rarest combination in AI startups: deep domain credibility in regulated finance, academic AI research, and the technical ability to build the platform himself.
Academic Background
Laurea cum Laude, University of Milan — Philosophy of Science, Epistemology. PhD Research at Harvard University (2003–2007) — Economics & Econometrics with John Nash and Gary Chamberlain.
Finance Career — 25 Years
CEO at Wealth Global Management (London). Head of Research at Banque Morval (Geneva). Fund Manager at Crédit Agricole (New York). Deep experience in the regulated industries PRISM serves.
AI & Technology
Professor of Machine Learning & AI, ITS Angelo Rizzoli — 15+ years teaching. Built PRISM's complete technical stack personally: infrastructure, backend, frontend, ML pipeline. Published research: arXiv 2311.16675. 5 registered AI patents (SIAE).
Track Record
Built €0 to €5M revenue businesses. Managed teams of 25 people. Published research on AI. The platform runs today — not a slide deck.
A Complete, Running Platform
Production system — multi-tenant, SSL-secured, deployed and operational. Commercial launch is the next step, not the current one.
Live Demo Available
Every investor conversation includes a live demonstration — a real agent, real documents, real style extraction, real reasoning. We can build a twin in your domain during the call.
app.aidometech.aiA Structurally Underserved Market
Regulated industries represent a majority of professional GDP in Europe — and they are systematically excluded from the AI tools everyone else uses.
| Capability | ChatGPT / Claude | Enterprise AI | PRISM |
|---|---|---|---|
| Data stays on your servers | ✗ | ⚠ Complex | ✓ |
| GDPR / HIPAA by design | ✗ | ⚠ Partial | ✓ |
| Writes in your voice | ✗ | ✗ | ✓ |
| Persistent memory across sessions | ⚠ Limited | ✗ | ✓ |
| Learns from your conversations | ✗ | ✗ | ✓ |
| Reasoning Layer — not just RAG | ✗ | ✗ | ✓ |
| Fixed predictable cost | ✗ Per token | ✗ €500K+ | ✓ |
| Setup in minutes, no AI expertise | ✓ | ✗ Months | ✓ |
Two Markets, One Infrastructure
Individual professionals (high margin, fast sales) and enterprise teams (higher revenue, stable contracts). Together they optimize utilization and reduce concentration risk.
B2C — Individual Professionals
- Each professional owns their own tenant
- Self-registration, fast onboarding
- Higher margin per seat
- First customers drive B2B referrals
- Target: Psychiatrists, Lawyers, Engineers, Advisors
B2B — Enterprise Teams
- Company pays one invoice, multiple departments
- Complete tenant isolation per department
- Dedicated onboarding and success management
- Higher total contract value, lower churn
- Target: Law Firms, Clinics, Financial Advisors
Real Numbers. No Optimistic Assumptions.
Infrastructure runs 24/7 regardless of usage. Every customer above breakeven is nearly pure margin.
| Component | Specification | Cost/mo |
|---|---|---|
| GPU Instance | L40S-1-48G — NVIDIA L40S 48GB VRAM | €1,022 |
| CPU Instance | PRO2-M — 16 vCPU, 64 GB RAM | €320 |
| PostgreSQL | DB-DEV-L — 4 vCPU, 8 GB, 100 GB SSD | €55 |
| Block + Object Storage | 200 GB SSD + 500 GB S3 | €27 |
| Networking & Monitoring | Static IPs, SSL, alerts | €31 |
| Web Search | Exa.ai semantic (primary) + Tavily (fallback) | ~€40 |
| Buffer (10%) | Unexpected costs + support tools | ~€155 |
| Total Monthly Cost | ~€1,950 |
| Scenario | Customer Mix | Revenue | Profit | Margin |
|---|---|---|---|---|
| Minimum | 5 Professional @ €249 | €1,245 | -€705 | Loss |
| Breakeven | 8 Professional @ €249 | €1,992 | +€42 | 2% |
| Target (Month 6) | 1 Business + 5 Professional | €3,244 | +€1,294 | 40% |
| Comfortable | 15 Professional | €3,735 | +€1,785 | 48% |
| Optimal (1 GPU) | 1 Business + 8 Professional | €3,991 | +€2,041 | 51% |
| Scale (2 GPUs) | 2 Business + 10 Professional | €6,488 | +€3,566 | 55% |
Path to €1M ARR in 36 Months
Conservative projections based on organic growth in two initial verticals. No viral assumptions.
Year 1
Year 2
Year 3
The Ask
A focused raise to fund commercial launch and reach Series A readiness. This raise takes PRISM to profitability — not just to the next fundraise.
Terms
Use of Funds
| Round | Timing | Raise | Pre-Money | Return on €250K |
|---|---|---|---|---|
| Pre-Seed (Now) | Q2 2026 | €250K | €1.5M | Entry |
| Series A | Q3 2027 | €2–3M | €10–15M | ~10x → €2.5M |
| Series B | Q1 2029 | TBD | €50–80M | ~40x → €10M+ |
"In five years, every professional will have an AI that thinks like them, knows what they know, and never forgets. PRISM will be the platform that makes this possible — privately, securely, worldwide."— Marco Gruppo, Founder & CEO
Schedule a Conversation
Every investor conversation includes a live platform demonstration. We can build a Professional AI Twin in your domain, with your documents, during the call.
Request a Conversation →
This document contains forward-looking statements. Actual results may differ materially.
This is not an offer to sell securities. Investment involves risk including potential loss of principal.