AI-Orchestrated Entrepreneurship Support Platform for Europe

European startups in emerging and moderate innovation ecosystems often struggle to find efficient support to adapt their project to their environment efficiently, scale, and overcome the lack of funding in later stages. Investors from innovation hubs likewise find it difficult to discover opportunities in less connected regions, hampered by insufficient information on local markets and regulations. Expanding investment ecosystems has become a priority to ensure all European regions can participate in and benefit from innovation-led growth.

The EU's "Expanding Investment Ecosystems (CONNECT)" call targets this gap by fostering structured programs that link strong and emerging innovation regions. Key recommended actions include creating central information points on local market conditions for foreign investors and startups, and compiling policy recommendations to ease cross-border investment barriers. The New European Innovation Agenda further urges initiatives to address "burning challenges" like mastering the digital transformation and achieving sustainability, ensuring that support services are tailored to pressing societal needs.

In response, MOSAIC Project proposes a novel solution to digitally orchestrate entrepreneurship support across borders, leveraging advanced Artificial Intelligence (AI) to connect and strengthen local ecosystems. The idea of this project was born from the construction of an intelligence model predicting success/survival/failure with a database of more than 40,000 entrepreneurs and an accuracy of 98%. Led by Pablo de Olavide University and Fundación Andalucía Emprende (the public regional entrepreneurship foundation of Andalusia, Spain), MOSAIC unites 8 partners in a 2-year, €1 million collaboration to develop and pilot an AI-based entrepreneurship support platform.

This platform will function as a modular "central point of information" and advisory system, providing entrepreneurs and startup support organizations with personalized guidance on markets, investment, and growth—while simultaneously facilitating foreign investor awareness of local opportunities. By combining cutting-edge AI (including multi-agent systems and large language models) with human mentorship from incubators across Europe, MOSAIC aligns with EU priorities to expand investment ecosystems and empower entrepreneurs in every region.

MOSAIC's core objectives are:

• Develop an AI-Orchestrated Support Platform: Build a modular entrepreneurship advisory system where specialized AI agents (as microservices) collaborate via a Model Context Protocol (MCP) to provide comprehensive, real-time guidance to startups. This includes advisory on market intelligence, business model validation, funding opportunities, and regulatory compliance.
• Pilot Across Diverse Ecosystems: Implement and pilot the platform in collaboration with entrepreneurship incubators and innovation hubs in multiple European regions (both "strong innovator" hubs and "moderate/emerging" ecosystems). Through at least 3 pilot sites, validate that the platform effectively supports startups in expanding to new markets and attracts interest from foreign investors.
• Human-in-the-Loop Empowerment: Integrate incubator staff and advisors as human-in-the-loop mentors who oversee and improve the AI agents. Train these professionals to adapt the AI services to local contexts, ensuring the advice is contextualized and trustworthy. This governance model will also foster cross-border mentoring, as experts from advanced ecosystems guide those in less developed ones.
• Enhance Cross-Border Investment Connectivity: Lower the information barriers between local startups and European investors. MOSAIC will serve as a one-stop information hub about local ecosystems (market conditions, incentives, regulations) for foreign investors, and conversely inform local startups about requirements in target markets.
• Ethical and Privacy-by-Design Innovation: Ensure the platform upholds European values—incorporating privacy-by-design (e.g. local/offline data processing options), transparency in AI decision-making, and compliance with emerging AI regulations. An ethics board and continuous evaluation will guide responsible AI use aligned with EU guidelines.
• Generate Evidence and Policy Insights: Rigorously monitor and evaluate the pilots to extract best practices and data-driven insights. MOSAIC will deliver a repository of case studies and policy recommendations for local and European authorities on how AI and cross-border collaboration can improve innovation ecosystems.

By achieving these objectives, MOSAIC will strengthen local and cross-border entrepreneurship support mechanisms, contributing to a more interconnected European innovation ecosystem and aligning with the Horizon Europe CONNECT call goals.

MOSAIC's platform is built on an innovative Multi-Agent System architecture underpinned by the Model Context Protocol (MCP). MCP serves as a common communication layer that allows multiple specialized AI agents to orchestrate their knowledge and actions seamlessly – akin to a "USB-C for AI" that standardizes how agents plug into the system.

Each agent is implemented as a microservice with a well-defined role, such as:

• Market Intelligence Agent: continuously gathers and analyzes real-time market data (industry trends, competitor analysis, customer insights) relevant to a startup's product.
• Business Model Validator: evaluates entrepreneurs' business plans or ideas against best-practice frameworks and historical success patterns (leveraging prior research like fuzzy Qualitative Comparative Analysis (fsQCA) modeling of startup success factors).
• Funding Advisor Agent: provides tailored information on funding opportunities – from EU and national grants to venture capital calls – and matches startups with potential investors or programs.
• Regulatory Compliance Agent: offers guidance on legal and regulatory requirements in target markets (e.g. incorporation, taxation, standards, data protection), helping startups navigate cross-border rules.

These agents collaborate under an Orchestrator Agent that manages the dialogue and context. For instance, when a user (entrepreneur or advisor) queries the system for scaling to a new country, the orchestrator might ask the Market Intelligence agent for market stats, the Regulatory agent for compliance checks, and the Funding agent for relevant investors/programs – then aggregate these into a coherent advisory output.

This modular microservice design means each AI agent can be updated or expanded independently (improving scalability and resilience), and new agents can plug in easily via MCP without complex re-integration. The project leverages previous developments by the partners, particularly a neural network-based entrepreneurial advisory system from earlier research that included fuzzy-set Qualitative Comparative Analysis (fsQCA) to capture combinations of success factors.

A distinguishing aspect of MOSAIC is the close interplay between AI agents and human expertise. Rather than replacing human advisors, the platform augments them. During development, entrepreneurship coaches and incubator staff from partner organizations will actively train and fine-tune the agents. We will implement a feedback mechanism where these experts can review the AI's recommendations and adjust parameters or provide corrections if necessary.

MOSAIC's design follows a strict ethics and privacy-by-design framework. All personal or sensitive business data processed by the platform will remain under user control, with options to opt out of cloud processing in favor of local on-premise runs. The AI agents will include transparency features (explainable AI techniques for key decisions) so that users understand why a recommendation was made.

The MOSAIC platform will be iteratively developed and piloted in real operational environments through a structured field trial:

Pilot Sites: We will deploy the platform in three European incubators or entrepreneurship centers, selected to represent a spectrum of ecosystems:

• Andalusia (Spain) – Led by Universidad Pablo de Olavide, representing a moderate innovator region with a broad base of start-ups, including many in traditional sectors undergoing digital transformation.
• Country B (Strong Innovator) – e.g., a partner in a highly developed startup hub (to be confirmed), providing access to cutting-edge start-ups and foreign investors. This site will help test the platform's ability to connect an innovation leader region with others.
• Country C (Emerging Ecosystem) – e.g., an incubator in Southeast Europe or an associated country, where resources are scarcer. This ensures the platform is validated in a less connected ecosystem and adapts to local needs.

Use Case Scenarios: At each pilot site, a common set of use cases will be trialed. For example, "Startup expanding to a new EU market" – the local entrepreneur will use the platform to get a checklist of regulatory steps and a list of potential investors in that market. Another scenario is "Investor exploring opportunities in region X" – where an investor persona uses the platform to learn about local startups and conditions.

Mentorship and Training: Each pilot will involve local incubator staff acting as tutors or facilitators. They will be trained in using the platform and will guide entrepreneurs in their cohort to interact with it. These staff will collect feedback from end-users and report back any issues or suggestions. Additionally, we will implement a cross-country mentorship exchange: staff from a more developed ecosystem will pair with those from a less developed one to compare how the platform performs and to jointly refine the AI agents.

Data Collection and Monitoring: Throughout the pilot, MOSAIC will collect usage data and outcomes in a privacy-compliant manner. Key metrics include: number of queries by entrepreneurs and their satisfaction rating; number of investor information requests and follow-ups; improvement in entrepreneurs' knowledge or readiness (assessed via pre/post surveys); and any actual cross-border engagements initiated as a result of the platform.

Evaluation: After an initial 6-month pilot cycle, the consortium will assess performance against expected outcomes. We will use surveys, interviews, and analysis of system logs to evaluate effectiveness, adoption and usability, accuracy and trust, and cross-border connectivity. Findings from this evaluation will inform a second iteration of the platform with improvements deployed for a shorter validation round.

By the conclusion of MOSAIC, we expect to achieve the following key results and impacts:

• AI Entrepreneurship Support Platform (Full Deployment): A fully functional MOSAIC platform deployed in pilot regions, combining the neural-network and rule-based advisory from previous work with the new multi-agent MCP architecture. This platform will be open-source (with modules released under an EU-friendly open license) to allow any innovation ecosystem to adopt or adapt it.
• Increased Startup Capacity for Expansion: Entrepreneurs who participate in the pilots will gain improved capacity to internationalize. Thanks to personalized, context-rich guidance, they will better understand regulatory hurdles, market entry strategies, and funding pathways in other European markets. We aim to document multiple cases of startups successfully using the platform to prepare expansion plans, find relevant programs/investors, or enter new markets faster.
• Greater Investor Awareness of Emerging Ecosystems: Through project outreach and platform use, foreign investors will become more aware of opportunities in the pilot regions. The platform's information hubs and project events (like a planned virtual "Investor meets Ecosystem" demo day) will showcase promising startups from less-known regions.
• Validated Multi-Agent AI Architecture: On a technical front, MOSAIC will demonstrate the viability of the MCP-based multi-agent approach in a real use case. We will document the architecture and performance in a technical white paper, contributing new knowledge to the AI community on orchestrating specialized agents for complex advisory tasks.
• Cross-Country Mentoring Toolkit: Based on the training and exchanges among incubator staff, we will develop a cross-country mentoring toolkit. This will be a practical guide for innovation support organizations on how to integrate AI tools into their services and how to collaborate across borders.
• Policy Recommendations and Best Practices: In line with the Horizon call expectations, MOSAIC will produce a repository of best practices and a Policy Recommendations report. The best practices will cover effective methods to facilitate market entry for both startups and investors, distilled from our pilots.
• Long-Term Sustainability and Uptake: We plan and commit to a Sustainability Roadmap as a deliverable, outlining how the MOSAIC platform and network will continue beyond the project. This includes potential funding sources or business models and integration into existing European platforms.

Collectively, these results will directly contribute to strengthening local and cross-border entrepreneurship policies and practices in Europe. By empowering even less-developed ecosystems with advanced yet accessible AI support, MOSAIC helps level the playing field – enabling talent from any region to find opportunities and investors across Europe. This aligns with the vision of a more cohesive European Innovation Area and will generate evidence and examples to fuel future innovation policy and programs.

The MOSAIC project brings together leading organizations and potential collaborators from across Europe who have been invited to participate in this innovative initiative. These partners and collaborators represent a diverse network of innovation ecosystems, entrepreneurship support organizations, and research institutions committed to advancing AI-driven entrepreneurship support across European regions.