Master's thesis · UC3M · June 2026
Compliance.AI
A business plan for EU AI Act compliance in SMEs
Prove your AI is legal, without hiring a lawyer.
- Programme
- Master in Management, Universidad Carlos III de Madrid
- Supervisor
- Professor Samira Reis
- Defended
- June 2026
- Grade
- 8.5 / 10
01/Abstract
Compliance.AI is a B2B SaaS platform designed to help European SMEs comply with the EU AI Act (Regulation EU 2024/1689). On 7 May 2026 the Council and the Parliament reached a provisional agreement under the Digital Omnibus to defer stand-alone Annex III high-risk obligations to 2 December 2027. The plan is built around that window.
It sets out the strategic positioning, market opportunity, competitive structure, product architecture and financial projections for a specialist compliance solution aimed at the gap left open by enterprise-focused competitors. The analysis draws on Porter's Five Forces, the resource-based view and organisational ecology, and positions Compliance.AI as a focused competitor in the SME compliance-technology market.
02/The plan in numbers
EUR 199
Per month for the Essentials tier, against a EUR 50k to 250k consulting bill
434,000
SMEs with 10 to 249 employees in the Year 1 markets: Spain, Italy, Portugal
700–1,200
Addressable firms after scope and reach filters
EUR 400k
Seed required, of which EUR 80k founder equity, no second raise
Month 33
EBITDA break-even from inception; net income positive from Year 3
60% → 82%
Gross margin, Year 1 to Year 5, as human review automates
03/Five-year financials
Base case. Revenue follows the customer count: 24 at the end of Year 1, 301 by Year 3, 843 by Year 5. Losses in the first two years, profit from Year 3, and the bank balance never goes below EUR 30,000.
| Y1 | Y2 | Y3 | Y4 | Y5 | |
|---|---|---|---|---|---|
| Revenue | 49,276 | 308,772 | 989,963 | 2,134,787 | 3,708,519 |
| EBITDA | (181,049) | (189,228) | 88,963 | 606,787 | 1,453,519 |
| Net income | (185,549) | (199,228) | 77,963 | 594,787 | 1,435,519 |
| Cash, end of year | 219,857 | 30,216 | 125,885 | 735,475 | 2,177,571 |
Year 1 starts at the August 2027 platform launch. No tax until Year 4: losses from Years 1 and 2 offset later profit, then Spain's 15% rate for new companies applies.
04/What the plan argues
- 01
The gap is structural, not temporary
Enterprise governance vendors price at USD 30,000 to 150,000 a year and cannot repay an enterprise sales team on a EUR 199 contract. Resource partitioning explains why they do not follow a specialist downmarket.
- 02
Four modules mapped to the Act
A risk-classification engine (Article 6, Annex III), an Annex IV documentation generator, a conformity workflow (Annex VI) and post-market monitoring (Article 72). Monitoring is where the switching costs build.
- 03
Three people, not ten
A founder on zero salary until EUR 240k ARR, a Head of AI Act Compliance and an Italy lead. European seed benchmarks run ten people on EUR 1 million; this plan runs three on EUR 400k.
- 04
Churn is the number to watch
The base case assumes 3.5% monthly churn. At 5% the lifetime value of an Essentials customer falls from EUR 3,411 to EUR 2,388 and the 3:1 LTV-to-CAC floor breaks once fully loaded acquisition cost passes EUR 800.
- 05
Exit and pivot triggers are written down
A sustained 25% fall in monthly customer additions, or a further six-month delay to the deadline, activates a focused exit to a GRC platform. Growth at 1.5 to 2 times base case pulls a Series A forward.
05/Structure
- 01Business idea and strategic positioning
- 02External analysis: PESTEL and Porter's Five Forces
- 03Internal analysis: VRIO, capability stack and SWOT
- 04Marketing plan: market sizing, pricing, four channels
- 05HR plan: the three-person team
- 06Operations
- 07Financial plan: five years, unit economics, scenarios
- 08Implementation plan
- 09Critical risks and contingency plan
Frameworks
- PESTEL
- Porter's Five Forces
- Resource-based view
- VRIO
- Organisational ecology
- Resource partitioning
- SWOT
- Unit economics
- Scenario and sensitivity analysis
91 pages · English
Sole author. The views are mine and were written as academic work, not as advice.