Bridging the Academia-Industry Gap by a Co-Design and Co-Research Approach for Generative AI Adoption

GenAI is transforming how individuals and businesses create, manage, and access knowledge, with numerous applications such as customer service, sales, marketing, and employee onboarding, to name a few. Despite growing global investment on GenAI, its effective integration remains challenging (Maslej et al. 2024). Selection and implementation of GenAI technologies depend on use cases, industry, AI readiness, and integration level, requiring specialized expertise that many SMEs lack.

The present project idea was driven by our research in applied AI and knowledge management. We decided to help companies develop and deploy GenAI solutions for knowledge access, capture, and creation through a business-oriented toolkit and support services. Insights from Haaga-Helia’s AI consultancy in the Finnish AI Region (FAIR) EDIH project highlighted challenges in GenAI adoption (Khan et al., 2025) and supported our project idea. To ensure the research result (the toolkit) is business-relevant, innovative and widely reusable, industry-university cooperation became essential.

ERDF Co-Research Funding

We selected the Innovation and Competence Networks National Theme Research Project Call 1/2024 (Innovaatio- ja osaamisverkostot valtakunnallisen teeman tutkimushankehaku 1/2024) as the most suitable funding instrument for our project, aligning with its focus on enhancing research and innovation capabilities and promoting the adoption of advanced technologies. This call supports initiatives that drive national innovation and strengthen competence networks, making it an ideal fit for the objectives and impact of our project.

To ensure competitiveness, new innovations arising from research and rapid commercialization of innovations are needed. The competencies of educational and research organizations should be used more effectively, faster, and more widely in companies’ business operations. The call intended to fund industry-academia collaboration by bringing together multiple businesses and several research organizations. Also, the whole project should be co-funded, fostering the real commitment of all participants.

Project Preparation Activities

The following project preparation activities were defined and executed:

1. Communication activities: Communication efforts were key to attracting businesses and broadening the project’s visibility and support. These included publishing awareness-raising articles such as (Khan et al., 2024) about using generative AI for knowledge management in SMEs. Another highlight was an awareness-raising seminar, “Generative AI-enhanced knowledge management in business,” featuring expert presentations, a panel discussion, and both in-person and online participation.
2. Project consortium formation: Building a strong consortium required engaging both companies and academic partners (at least 3 companies and 2 academic partners). Companies, as potential end-users of the generative AI toolkit, were crucial for co-financing and testing project solutions, while academic partners brought complementary expertise in Large Language Models (LLMs), knowledge management, and change management. Helsinki University and Tampere University, our academic partners in the consortium, helped refine the project vision and proposal. This collaboration ensured a well-rounded consortium supported by letters of interest from industrial partners.
3. Defining use cases and impact: The articulation of use cases involved two steps: formulating generic GenAI use cases from a literature review and linking them to specific examples in several application areas (e.g., sales, customer service), followed by refining and prioritizing these use cases through empirical analysis of companies’ needs. This process included business dialogues, input from companies’ experts, and co-creation methods. Further details are available in the KMIS 2024 conference paper (Kudryavtsev et al., 2024).
4. RDI challenges identification: The identification of the RDI challenge focused on complex issues that required research and development to solve. This process involved a literature review, a panel discussion with experts at the awareness-raising event, and co-creation workshops with academic partners and technology providers.
5. RDI project planning: The comprehensive project plan was developed iteratively. The Haaga-Helia team prepared an initial draft detailing project objectives, deliverables, work packages, timeline, and responsibilities. This draft was further refined and expanded through discussions with companies and academic partners. The project plan is based on the principles of action design science research (Sein et al, 2011) – we plan to iteratively develop IT-based artefacts, implement them in companies, get feedback and elaborate them.
6. Funding application: This activity was primarily technical since we prepared all the content during the previous activities.

All these activities were supported by the preparatory ERDF funding. We executed the aforementioned activities, and submitted the project application in mid-2024, which was successfully approved by the end of the year. Now we launched the project – Generative AI-Enhanced Knowledge Management in Business (GAIK) .

Lessons Learned

Throughout the project preparation, we learned several lessons that enhanced our approach to collaborative research and development. Some of these lessons are mentioned below.

A) Value proposition to companies: Companies want to understand how research outcomes will benefit their operations, improve efficiency, or solve specific challenges. This highlights the need to frame research proposals with a clear value proposition relevant to their business. Early involvement of companies in project design is important to co-develop approaches that meet all participants’ needs.

B) Communication strategies for company engagement: Another significant challenge we encountered was the difficulty in persuading companies to commit to a future solution, particularly given the limited time that company professionals could dedicate to these discussions. Persuasive slides and letters are essential but not sufficient. One-to-one personal contacts with relevant persons in companies are the most effective way to get companies on board and build the consortium.

C) Balancing industry needs and research goals: Many companies expected a ready-made software solution, while universities are interested in research results that are available to the public, generalized, reusable, and innovative, addressing typical business problems. To bridge this gap, the project aimed to deliver both research-oriented outputs (e.g., a reusable GenAI toolkit) and company-specific solutions (e.g., tailored applications enabling toolkit testing). Software development partners were included in the project proposal to enhance the technology readiness of these solutions. Figure 1 illustrates the use cases (problem space) and expected project results (solution space), highlighting the project’s dual focus on generic and company-specific outcomes.

D) Breaking project implementation into iterations is highly valuable: This iterative approach enables companies to receive an initial simple solution within months and gradually improve it with advanced versions that address limitations, incorporate user feedback, and integrate research findings.

E) Addressing data privacy concerns: Data privacy was a major concern for many companies, particularly regarding data handling and protection. Additionally, they questioned the accuracy and trustworthiness of AI-generated outputs. To address these concerns beyond standard NDAs, we emphasized the safeguards and protocols that would be implemented to ensure data privacy and protection and the measures we would take to ensure the accuracy and reliability of the AI-driven solutions.

F) Addressing technical expertise gaps: An important challenge was the limited technical expertise among company professionals, which led to hesitation in fully committing to the project. To address this, we simplified our communication, focusing on core benefits and applications of generative AI, and offered additional consultancy and requirement analysis sessions to build confidence and engagement.

So, how can the quality and impact of RDI projects be improved by following a similar co-design and participatory approach to project preparation and planning? We see the following combination of benefits for universities:

1. Co-financing and active involvement from industry partners, leading to more impactful results;
2. Opportunities for academic contributions, including peer-reviewed publications, beyond basic software development or consultancy in the interest of companies;
3. More realistic cooperation expectations from all sides, reducing the risk of conflicts and dissatisfaction during the project;
4. Better alignment between the companies’ needs and capabilities of universities that reduce the risk of project failure;
5. Faster project start since companies’ use cases were already defined – more time for RDI activities.

The word ”combination” is important — some benefits can be achieved with simple approaches but at the expense of losing other benefits.

Conclusion

The industry-academia co-research project is a fascinating and motivating approach to carry out R&D. It stems from real practical problems and simultaneously brings academic results. Such collaborative working means learning together – practitioners learn from each other and from researchers, and research teams gain more practical insights.

This article demonstrated a possible approach for establishing such cooperation and highlighted existing challenges and ways to address them. The expected value from such a participatory approach to project preparation was defined.

Picture: AdobeStock_industry_university_cooperation

LISÄÄ AIHEEN YMPÄRILTÄ / RELATED POSTS