AI R&D for European Commission Funding: AI Act Challenges
R&D Projects in the Era of AI Growth
In today’s fast-evolving technological landscape, the motivation for initiating R&D projects has never been stronger. Artificial Intelligence (AI) technologies are advancing at a rapid pace, driving innovation across various sectors—from healthcare to agriculture and manufacturing. This acceleration has made AI not only more accessible to businesses of all sizes but also easier to develop and integrate into new systems and workflows. Companies can leverage cutting-edge AI tools to solve complex problems more efficiently than ever before.
However, with these newfound capabilities, there comes a need for structured planning, especially when venturing into novel research domains. While AI tools can automate certain processes, successful R&D still requires careful, long-term planning to ensure outcomes align with strategic goals. The development of innovative AI systems demands rigorous experimentation, hypothesis testing, and prototype creation. In this context, research and development (R&D) projects must be thoughtfully planned, especially when applying for European Commission funding. This planning ensures that AI- driven innovations are not only technologically sound but also capable of meeting the needs of evolving markets.
The European Commission offers a range of funding opportunities, particularly for AI-related innovations, but the successful execution of these projects hinges on careful preparation. With the increasing push towards AI-driven solutions, companies must align their R&D projects with both technical and regulatory demands, considering the growing legal framework surrounding AI deployment. Thus, R&D project planning is essential to not only advance technological innovation but also to stay competitive in an AI-powered world.
R&D Projects and Technology Preparedness Levels (TPL)
| TRL | Description |
| TRL 9 | Actual AI system deployed and proven in real-world operations (e.g., fully operational AI in production) |
| TRL 8 | AI system completed and tested in an operational environment (e.g., final validation and qualification in real use cases) |
| TRL 7 | Prototype AI system demonstrated in an operational environment (e.g., tested with real-world data in the target environment) |
| TRL 6 | AI system or subsystem prototype demonstrated in a relevant environment (e.g., system is tested with representative data) |
| TRL 5 | AI component or algorithm validation in a relevant environment (e.g., tested with real or near-real data) |
| TRL 4 | AI component or algorithm validation in a laboratory environment (e.g., validated in a controlled setting with simulated data) |
| TRL 3 | Experimental proof-of-concept for AI system (e.g., algorithms validated through initial testing or simulations) |
| TRL 2 | AI concept and initial methodology formulated (e.g., theoretical models for AI approach are developed) |
| TRL 1 | Basic principles and underlying AI technologies observed and reported (e.g., initial research and AI model exploration) |
Table 1. Technology Readiness Levels (TRL) Framework
When planning R&D projects, especially those seeking European Commission funding, it is crucial to align the project with the Technology Preparedness Levels (TPL). These levels, ranging from 1 to 9, provide a structured pathway from initial research and discovery to the commercialization of a product. Each level represents a stage of development, starting with basic scientific exploration and culminating in the deployment of fully operational systems in real environments. Below, we break down each of the nine TPL stages, providing examples for clarity.
- TPL 1 – Basic principles observed. At this stage, foundational scientific knowledge is acquired to establish the underlying concepts of the technology. In the context of AI, this might involve identifying key challenges or exploring theoretical frameworks, such as mathematical models for AI-driven agent systems. This stage focuses on understanding the basic principles without any practical or experimental implementation, laying the groundwork for future research and development.
- TPL 2 – Technology concept formulated. Here, theoretical concepts begin to take shape. In the case of a smart house management system, a theoretical model that integrates AI-based control systems to manage the infrastructure of a multi-apartment building might be conceptualized. The focus is on defining the essential functional relationships and identifying critical design challenges.
- TPL 3 – Experimental proof of concept. The theoretical model is further refined and validated in a controlled environment. For example, in a smart house system, the initial functionality for controlling temperature or lighting could be tested through simulations or small-scale experimental setups.
- TPL 4 – Technology validated in lab. This stage involves creating a prototype of the system. For instance, an alpha version of the smart house management IT system would be developed and tested under lab conditions, with key features such as data flows and information encoding algorithms tested for efficiency and functionality.
- TPL 5 – Technology validated in relevant environment. Moving beyond the lab, the system is validated in a real-world environment. In our smart house example, this could involve deploying the system in a pilot residential building to ensure that the AI functions as expected when exposed to real-world data and conditions.
- TPL 6 – Technology demonstrated in relevant environment. This stage involves demonstrating the full system in the relevant environment, with prototypes performing their intended tasks. The beta version of the smart house system would be equipped with all necessary configurations and tested under various conditions, making any needed adjustments based on real-time feedback.
- TPL 7 – System prototype demonstration in operational environment. The smart house system, now in its final prototype form, would be demonstrated in a real operational environment. It would connect to existing infrastructure in the building, adjusting for any specific conditions, and proving that the system can reliably manage the building’s operations.
- TPL 8 – System complete and qualified. The system is now complete, with all testing and adjustments made. It is ready for deployment in environments closely matching potential customer requirements. Documentation is prepared, including user manuals, installation instructions, and performance data. The smart house system would undergo rigorous testing across different seasonal cycles to ensure its reliability and quality.
- TPL 9 – Actual system proven in operational environment. At this final stage, the smart house system is commercially launched and fully operational. It has been validated and tested, and it is now available for market deployment, complete with recommendations for installation and evidence of its economic efficiency.
By following these TPL stages, companies can methodically advance their R&D projects from ideation to market. For example, this structure is relevant not just for smart house systems but also for AI-driven technologies such as agentic workflows using LLMs (large language models) for remote sensing analysis via satellite APIs. In both cases, research and development must be planned with a clear understanding of the path from basic concept to full system deployment. Proper alignment with the TPL framework helps ensure that a company’s innovation is on track to meet both technical and commercial objectives.
Frascati Manual Criteria for R&D Project Evaluation
To increase the likelihood of winning funding from the European Commission for R&D projects, companies must ensure that their projects meet the rigorous standards set out in the Frascati Manual. This manual defines the criteria used to evaluate research activities and determines whether they qualify as R&D. Projects must fulfill four key criteria, which assess the originality, knowledge creation, uncertainty, and systematic nature of the research.
- New Knowledge Creation The first criterion requires that the project aims to generate new or additional knowledge. This could involve addressing scientific or technological problems that have not yet been solved or applied in the specific industry or service sector. For example, if the project involves AI-based solutions for remote sensing, the key is that the knowledge needed to solve the problem is not publicly available or widely applied in the sector. The outcome of the project should either introduce new products, processes, or significantly improve existing ones.
- Original Ideas or Hypotheses Projects must be based on original ideas or hypotheses. For instance, if an AI project is trying to solve a specific problem using large language models (LLMs), the project should explore innovative ways to improve or develop new workflows for satellite data analysis. These original ideas should provide a solution to the scientific or technological challenges the project is addressing. Methods like literature reviews, experiments, or surveys should add significant value by creating new knowledge that contributes to the project’s practical outcomes.
- Uncertainty A crucial aspect of any R&D project is the presence of uncertainty. There should be a recognized probability that the project might not yield the expected results. This could include the risk of not obtaining new knowledge of sufficient quality or quantity, failing to achieve the planned outcomes within the allocated budget, or encountering challenges in meeting the goals within the project’s timeframe. For instance, when developing AI models for satellite analysis, there is always a chance that the models will not perform as expected due to unforeseen variables or limitations in the data.
- Systematic Approach The project must follow a systematic, logical structure in its activities, ensuring consistency in planning and execution. This includes adhering to the SMART principles—Specific, Measurable, Achievable, Relevant, and Timed. The R&D activities should also align with the recommended stages of research and development, as outlined in the project plan. Furthermore, the documentation of the project should allow for the results to be replicated and passed on, ensuring the knowledge can be used by other researchers or applied in future projects.
By adhering to these four criteria—creating new knowledge, being based on original hypotheses, addressing uncertainty, and following a systematic approach—companies can significantly improve their chances of securing European Commission funding for their R&D projects. In addition to meeting the core Frascati Manual criteria, it is crucial for companies to conduct a thorough analysis of the existing scientific literature, patents, and commercial products before formulating their project approach. This ensures that the proposed solution is genuinely innovative and not a reiteration of existing technologies. Very often, experts evaluating project proposals conclude that similar solutions already exist, resulting in the project being categorized as technical work rather than true R&D. In such cases, funding is typically not provided. To avoid this, companies must clearly demonstrate how their project builds on current knowledge and offers a novel contribution, ensuring it qualifies as genuine research and development.
Additional Considerations for R&D Project Planning
When applying for European Commission funding, there are several additional issues companies must carefully consider to ensure successful project execution and funding acquisition.
Funding is a Subsidy, Not Full Capital. While the European Commission provides subsidies to support R&D projects, it’s essential to remember that these funds typically cover only a portion of the project’s overall cost. Companies must have access to private capital or other sources of funding to cover the remaining expenses. Although the European Commission may provide an advanced payment to get the project started, this alone is insufficient for the entire project lifecycle. Careful financial planning is required to ensure the company can sustain the project even before the grant is awarded.
Long Evaluation Period for Project Proposals. The evaluation process for R&D funding proposals can be lengthy, sometimes taking several months. If a project proposal is rejected, the resubmission process can add further delays. This timeline can be a significant challenge, especially in the rapidly evolving AI sector, where market conditions may change before the project even begins. Therefore, companies should plan to invest from private capital to start initial project activities and continue development in parallel with the funding application process. This approach helps avoid a slowdown in project momentum in case of a delayed or rejected application.
Planning for the AI Act and Market Commercialization. With the upcoming AI Act, companies must consider how their projects align with evolving regulations. The AI Act emphasizes the importance of developing AI systems that are ready for commercialization, especially in Technology Preparedness Levels (TPL) 7-9. However, this can create tension with the earlier stages of research (TPL 1-6), where experimentation and exploration dominate, and the final market fit might still be uncertain. It can be challenging to plan commercialization activities when the project is still in its initial phases of research. To mitigate this, companies may need to seek external consultations with legal and market experts to ensure that their R&D efforts are not only aligned with regulatory requirements but also adaptable to future commercialization efforts.
Navigating Market Uncertainty. In many cases, especially with AI projects, the market fit for new technologies may not be immediately clear. This uncertainty adds complexity to project planning, as it becomes difficult to anticipate how the AI system will be received or what specific commercialization strategies will work. To address this issue, companies should engage in continuous market analysis throughout the R&D process. External consultations with industry experts can help companies better plan for market integration, even when dealing with uncertain market conditions.
By taking these additional considerations into account, companies can better manage the inherent challenges of R&D projects, particularly when seeking European Commission funding. Thorough financial planning, strategic project timelines, and compliance with evolving AI regulations will all contribute to the success and sustainability of R&D initiatives.
Reference:
- Lietuvos inovacijų centras Mokslo inovacijų ir technologijų agentūra, Mokslinių tyrimų ir eksperimentinės plėtros išlaidų apskaitos vedimas ir MTEP projektų dokumentavimas REKOMENDACIJOS, 2017
- Organisation for Economic Co-operation and Development (OECD), Frascati Manual: Guidelines for Collecting and Reporting Data on Research and Experimental Development, 2015
- Organisation for Economic Co-operation and Development (OECD), Oslo Manual: Guidelines For Collecting, Reporting And Using Data On Innovation, 2018
- Petraitė, M., & Kucinas, R. (2014). MTEP lygių apibrėžčių metodologija.
About author
Dr. Valentas Gružauskas is an expert in AI governance, conformity assessment, and ethical AI applications. As an Associate Professor at Vilnius University and CEO of AI Conformity & Research Consulting, he leads initiatives in AI regulation and ISO standards. He actively participates in AI governance through the Horizon Europe Programme and various standardization committees. Dr. Gružauskas has made significant contributions to fields like machine learning, deep learning, and agent-based modeling, with a focus on AI ethics and data depersonalization. His work is widely published in esteemed journals, reflecting his impact and recognition in the AI community.