AI Transformation Award Finalist · Malt
AI Transformation Award · Malt
🥉 Third place

The competition was incredibly tight among the 34 finalist projects, out of a total of 161 submissions. The winners stood out by a very small margin, driven by their industrialization, measurable ROI, and the strong support from the public vote.

"Exceptional across every dimension. The AI-accelerated data governance transformation is both technically rigorous (400+ documented assets, 90% ADR coverage) and culturally sensitive (emotional intelligence in change management). The 4x onboarding improvement and 60% governance adoption demonstrate real transformation."

Galde's project was one of the standout highlights of this edition, and being a finalist is a significant achievement in such a competitive cohort.

Visit the case →
Logo de Galde InfoJobs · Adevinta — España

Data governance with generative AI on Spain's largest job platform

InfoJobs, Spain’s leading job board, faced a complex challenge: scaling its data maturity without disrupting operations or generating internal resistance. What we did was different: we used generative AI not only as a technical tool, but as a lever for adoption and organizational change.

Data Governance Generative AI dbt Elementary Change Management
-80%

documentation
time

faster
onboarding

400+

documented
assets

€80K

projected anual
savings

75%

autonomous team
after the project

The project

InfoJobs (part of the Adevinta group) had grown exponentially. Its data infrastructure had scaled at the same rate, but governance, documentation, and standardized practices had lagged behind.

The timing was especially critical: InfoJobs was about to become an independent legal entity, which required translating Adevinta’s global governance framework to its own operational reality without disruption.

The problem: 4 critical pain points

1. Lack of strategic clarity

Multiple data initiatives were running in parallel without a unified vision or prioritization framework. Teams didn’t know which governance practices to adopt first or how to sequence implementation.

2. Documentation debt

Años de crecimiento rápido habían creado lagunas significativas en la documentación de Redshift y Databricks. El conocimiento tribal dominaba, generando cuellos de botella y puntos únicos de fallo en el onboarding de nuevos miembros del equipo.

3. Resistance to change

Previous attempts to implement governance had failed due to poor change management. Teams perceived governance as bureaucracy rather than a value-adding structure, creating passive resistance to the new processes.

4. Technical-business gap

Data engineering and business stakeholders spoke different languages. Technical solutions were proposed without fully considering organizational dynamics, while business requirements were not translated into actionable technical data roadmaps.

The solution: Generative AI at the service of organizational change

The differentiating factor was not the technology itself, but the methodology. Instead of imposing a top-down technical framework, the approach was to use generative AI as a lever for adoption and knowledge transfer.

AI-generated educational content

Customized onboarding materials, tailored to each role and technical level, were generated using Claude and Gemini Pro to accelerate the adoption of dbt + Elementary. These materials included interactive tutorials, FAQs, and troubleshooting guides adapted to each user profile.

Automated documentation

Workflows assisted by LLMs analyzed Redshift and Databricks schemas, generated draft documentation that engineers refined, and validated consistency across assets. Result: an 80% reduction in documentation time.

Communication by stakeholder profile

Analysis of JIRA tickets and past communications to identify patterns in successful versus failed change initiatives. Generation of customized communication strategies for each group and governance impact simulators to visualize ROI in their specific context.

Foundational technical assets

Beyond the AI ​​layer, the technical foundations of the framework were delivered:

  • Formalized Architectural Decision Records (ADRs) for decisions from 2025 onward.
  • Golden Path 2025-26 aligned with InfoJobs’ strategic objectives.
  • dbt integration roadmap with quarterly milestones.
  • Complete documentation for Redshift and Databricks.
  • Initial definition of Claude Code skills for the internal team, usable independently.

Technology stack

Data Governance: dbt, Elementary,  Unity Catalog

Generative AI Layer: Claude API (templates and skills generation), Gemini Pro (documentation analysis)

Processing and Storage: Redshift, Databricks

Change Management: Confluence, JIRA (with semantic search and autogenerated summaries)

Measurable Impact

Operational efficiency

  • 80% reduction in documentation time: from 4 hours to less than 1 hour per pipeline.
  • 90% faster ADR creation: from 2+ days to same-day completion.
  • 4x faster onboarding: from 8 weeks to 2 weeks to full productivity.

Framework quality

  • Over 400 documented data assets, compared to less than 30% previously with adequate documentation.
  • 90% ADR coverage for relevant 2025 architectural decisions.
  • 60% adoption of dbt + Elementary in the pilot phase within established processes.

Strategic value

  • An 18-month roadmap with prioritized initiatives and quarterly milestones.
  • 75% of the team is confident in continuing the practices autonomously after the project (compared to the typical 40–50% in other external consulting projects).
  • Estimated 20% reduction in data incident resolution time.
  • Projected annual savings of €80,000 through infrastructure optimization and improved observability.
  • The methodology has been studied and adopted as a prototype for replication in other legal entities within the Adevinta group.

Why this approach works where others fail

The main reason governance initiatives fail isn’t technical: it’s organizational resistance. This project demonstrates that when AI is used to eliminate friction in knowledge work, rather than to replace human judgment, adoption accelerates dramatically.

The framework didn’t just gather dust. Two months after the project’s closure, the internal team was still executing the defined roadmap, using AI tools independently and accumulating over 400 documented assets that continue to grow in value.

Generative AI isn’t a substitute for data expertise. It’s the layer that enables that expertise to be transferred faster, farther, and with less friction.

Does your company face similar challenges?

If your organization is scaling its data infrastructure but governance, documentation, or adoption of best practices are not keeping pace, we can help you define a concrete plan.

Discover the rest of our services for businesses