Seeds of Change: What a Hackathon on AI and Vulnerability Reveals on Innovation Policy

Dr. Begoña G. Otero, Coordinator Group 1.4 AI4Good, Project cAIre

In February 2025 OdiseIA held the OdiseIA4Good Hackathon. It was neither a technology showcase nor a classic incubator. Designed as part of the cAIre research project, with the support of Google.org and the Pablo VI Foundation, and with the collaboration of entities such as the CEU San Pablo University, the ONCE Foundation, Innova-tsn, the GDG Madrid community, and other allies, the event was conceived not as a startup competition, but as a laboratory of possibilities. A testing ground to explore what kind of AI can emerge when the starting point is not efficiency, but vulnerability.

During those three days, more than 300 participants in 42 teams, ranging from experts and students in technology, ethics or data, to activists from various sectors, worked together to address real challenges related to the employment of people with disabilities, migration, the digital divide and other social, climate, educational, etc. issues. All of them with a global dimension and all of them focused on vulnerability. The dynamics combined technical workshops, specialized mentoring and final presentations before a multidisciplinary jury. The honorary presidency of the event was accepted by His Majesty the King of Spain, underlining its institutional and ethical relevance. The atmosphere was an unusual combination of collaborative creativity, technical rigor and social commitment, with a clear focus: to advance forms of AI that are not only technically feasible, but also centered on human dignity.

Five months later, the first signs of growth are visible, quiet, persistent, and unconventional. Most of the seeds planted during the Hackathon have not taken root in traditional innovation labs or startup accelerators. Instead, they are sprouting in places where urgency meets care: within community networks, cultural initiatives, informal collectives, and everyday problem-solving strategies. What is taking shape are not polished market-ready products, but living processes, adaptive, grounded, and deeply human. These efforts challenge the dominant narrative of “fail fast, scale fast” and instead claim their own pace, logic, and legitimacy. In this slower rhythm, innovation is not a sprint but a cultivation, and what blooms may not fit the mold, but it matters all the more.

What We’ve Observed

Not all of the 42 original projects continued. Some remained at the prototype stage, while others, as often happens in open calls, did not respond to follow-up. But six living and diverse trajectories offer something more valuable than polished products: persistent, evolving processes. These six teams, each in their own way, are still exploring how to turn early insights into long-term solutions.

Their scope is wide. Woowii is developing a tool to foster employment based on affinity, tackling recruitment bias. WeRepresent automates intelligent lead generation for underrepresented artists and athletes. Opills pivoted toward healthcare, building an app to manage medication in dependency settings. ElderCare Connect works on accessible robotic tools for daily care. Migration and Asylum Assistant offers multilingual support for navigating migration procedures. And Tramits-Easy, which emerged from a corporate environment, developed a prototype interface to facilitate digital administrative procedures.

These six projects have taken very different paths: some have advanced with strategic clarity, others have paused or are redefining their path. Yet, what stands out the most is not their degree of progress but the shared patterns that allows us to identify common trends in the real challenges faced by this kind of initiatives.

One of the key challenges is the absence of legal or technical structures. Several of the teams lack a formal legal entity, which limits their ability to access funding, sign agreements or piloting their solutions in real environments. In many cases, they also lack stable technical roles, limiting their ability to maintain or further develop their prototypes over time.

Secondly, they share the mismatch between the support offered and real needs. Available mentoring frameworks are designed for startups seeking scalability or investment. However, several of the projects supported do not pursue this model. Some are driven by students, artistic collectives or independent professionals seeking impact, not venture capital.

A third common trend is the lack of validation spaces. Even teams with functional prototypes find it difficult to access testing environments, including residencies, administrations, and vulnerable groups. Without these connections, even the most technologically viable projects stagnate.

Yet, the most striking pattern we have observed is that teams not only design for vulnerability, but from it. They operate without stable funding, without legal structures, with little formal support, and under significant personal strain. This dual condition, creating solutions for vulnerable contexts while working from precarious ones, shapes both the ethical and strategic character of the support they need.

Human-centered AI

Our reading is deliberate, not incidental. The fact that the six projects are not only designed for vulnerability but also emerge from it resonates with the theoretical framework we developed in OdiseIA Group 1.4, as outlined in our article "AI for Good: Questioning the idea of human vulnerability". There, we argue for moving beyond the notion of vulnerability as a trait of “the other”, something externally diagnosed or passively endured. Instead, we conceptualize it as structural, relational, and shared, a condition that shapes both designers and users alike. What matters is not who is vulnerable, but how we design technologies that either alleviate or exacerbate that vulnerability.

From this perspective, the six accompanying projects offer concrete examples:

• Reduce cognitive and administrative burdens for migrants and the elderly.

• Digital agents designed to help independent artist access cultural opportunities for artists without the back up of a formal agency.

• Explore more inclusive job selection models, where human affinity takes precedence over algorithmic efficiency.

• Facilitate medical self-care by adapting solutions to institutional contexts such as nursing homes.

  
  

In all cases, AI appears not as an end in itself, but as a means to a concrete and legitimate human goal. Its value lies not in the sophistication of the model, but in its adjustment to the context, its adaptation to the real capabilities of the users, and its potential to lower barriers. This is the basis of a truly situated AI.

Funding Ideas, Not Structures: A Critical Reading of Horizon’s Logic from the IEP-ifo report and OdiseIA4Good learnings

The follow-up of OdiseIA4Good projects is not just a one-off X-ray. When compared with the structural recommendations of the recent report Funding Ideas, Not Companies. Rethinking EU Innovation Policy From the Bottom Up (IEP-ifo), profound coincidences emerge about the current limits of public innovation policies in Europe. This report, prepared by the Institute for European Policymaking and the ifo Institute, offers a structural critique of EU’s current public innovation policies. Its core argument is clear: the existing system places disproportionate emphasis on funding companies that conform to conventional business models. As a result, it tends to exclude, or actively discourage, innovation emerging from informal contexts, social collectives, cultural networks, or grassroots initiatives. In short, the prevailing approach funds companies, not ideas, and systematically penalizes non-traditional forms of entrepreneurship and context-driven innovation.

Instead of prioritizing immediate economic growth or "market success," the report proposes redirecting public funding toward ideas - especially those with social, experimental or community value - even when they are not yet legally formalized.

Five aspects stand out in particular:

1. Funding Ideas Without Requiring a Legal Form

The IEP-ifo report argues that many ideas of public value are born outside formal structures, and that the requirement to be legally constituted blocks their early development. In OdiseIA4Good, this barrier was visible: several teams have not been able to move forward because they lack the legal entity to receive funding, enter into agreements or access institutional pilots. The report recommends the creation of direct access funds for pre-market ideas.

2. Reducing Bureaucracy and Simplifying Funding Access

The IEP-ifo report stresses that current administrative processes favor large players who can assume compliance costs or justify activities with professional means. OdiseIA4Good projects, many of them driven by students, precarious collectives or cultural entities, do not have this capacity. Bureaucracy becomes an obstacle rather than a guarantee. Debureaucratization is not optional: it democratizes access to innovation.

3. Contextual Support over One-Size-Fits-All Mentoring

One of the findings of our monitoring is the mismatch between the support offered and the real needs of the teams. Most of them are not looking to set up a startup or pitch to investors. Instead, they require legal guidance, access to institutional networks or support in scraping, backend or pilot testing. The IEP-ifo report suggests moving from mass acceleration programs to flexible support structures, tailored to the type and maturity of the project, and recognizes that specialized mentoring is most effective when it responds to the logic of the project, not the market mold.

4. Measuring Impact Beyond Return on Investment

The report also highlights the need to incorporate metrics of public value, social impact, inclusion and sustainability as evaluation criteria. The Hackathon experience confirms this: many of the projects analyzed do not promise profitability, but do promise dignity, accessibility and representation. Broadening the concept of innovation—from purely technological outputs to social, organizational, and situated processes—requires public funders to recognize relational, ethical, and contextual dimensions as integral components of innovative capital.

5. Recognizing the Value of Innovation Emerging from the Margins

Finally, the IEP-ifo report emphasizes that the ideas with the greatest transformative potential are not always born in academic or business environments, but rather on the margins: migrant communities, artistic networks, volunteer teams or underfunded collectives. OdiseIA4Good has eloquently reflected this. The fact that some of the most significant prototypes have arosen from positions of organizational vulnerability, without stable support or business structure, validates this thesis. We need policies that recognize this innovation not as anecdotal, but as strategic.

In this sense, the IEP-ifo study lends solid theoretical and policy legitimacy to the approach behind Odiseia's cAIRe project and its AI4Good framework: innovation focused on vulnerability, social justice or the public good cannot be assessed or supported using the same instruments designed for commercial entrepreneurship.

For Those Who Can Water the Seeds: Redesign Over Scaling-Up

Public administrations, universities, foundations and private entities not only have financial capacity. They also shape the rules of the game: they define what is considered innovation, who can participate and under what conditions. They are the ones who can turn the margins into seedbeds if they decide to water where institutional support does not currently reach.

The proposal is not to increase public spending but to rethink its architecture. Many of the obstacles faced by teams do not stem from a lack of funds but from their exclusive or poorly contextualized design. To foster plural forms of innovation - capable of encompassing both startups and non-profit collectives - it is essential to:

• Launch open calls and support schemes for early-stage initiatives, even if they lack a formal legal structure.

• Establish testing environments within public institutions, schools, healthcare centers and cultural venues.

• Provide technical, legal, and ethical support networks that go beyond traditional investment models

• Introduce and implement impact indicators that value collaboration, sustainability and contextual relevance, not just economic projections.

  
  

Some of OdiseIA4Good's teams compete for investment rounds. Others never will. Some were already companies before the Hackathon. Others do not yet have a legal status. What unites them is not their institutional form, but their potential to address real-world challenges through situated, thoughtful, and purpose-driven technologies. It is time for public policies and innovation ecosystem actors to recognize this plurality—not as an anomaly, but as a strategic asset.

We’re Still Planting

This report does not close a chapter: it’s part of the ongoing cAIre research project, running through early 2026. Our commitment remains: to document, support, and learn from socially driven AI projects born from structural vulnerability.

Throughout the project, we will continue to accompany the teams from a research perspective, connecting learning, identifying shared barriers, and proposing forms of intervention adapted to unconventional trajectories.

Out of respect for the teams, we do not disclose here confidential details or data that could compromise their future developments. Visibility is important, but it must be built carefully, without exposing those who are still in fragile processes.

The seeds are there. And from cAIre and OdiseIA, we will continue to observe, learn and prepare the ground so that they can flourish.

Bibliography

• Clemens Fuest et al., Funding Ideas, Not Companies: Rethinking EU Innovation Policy from the Bottom Up, Institute for European Policymaking & ifo Institute (2025). Disponible en: https://www.econpol.eu/publications/funding-ideas-not-companies

• Begoña G. Otero, Grupo 1.4 de OdiseIA, AI for Good: la idea de la vulnerabilidad humana en tela de juicio, OdiseIA Blog (2024). Disponible en: https://www.odiseia.org/en/post/ai-for-good-la-idea-de-la-vulnerabilidad-humana-en-tela-de-juicio