CASE STUDIES

What we've taken into production

Real projects in companies and public administrations: from business challenge to result, with data and AI in their place. Filter by area to see the ones that interest you.

Furniture and interior design company

AI furniture design with automatic quoting

Challenge: Customers could not picture how the furniture would look in their home, and every interior design proposal and its quote was prepared by hand, with days of waiting before they had a clear idea of the result and the price.
Solution: An AI interior designer embedded in their website with FastRender: the customer uploads a photo of the room, picks pieces from the catalogue or asks for a set matched to their budget, and the AI generates several proposals on their own photo, which they can adjust with “Request a change”. Each design works out the list of pieces and a preliminary quote with the catalogue prices.
Result: Customers see their space furnished with real catalogue products and a price estimate in minutes, and the sales team receives the opportunity with the photo, the design and the quote ready to close.

FastRender
Generative image AI
Furniture catalogue
Quote calculation

FastRender

Diversified business holding

From SAP to a lakehouse in Microsoft Fabric

Challenge: Reports spread across SAP R/3, SAP BusinessObjects, several ERPs and CRMs. Without a single analytics environment, there was no way to add advanced analytics or contain licensing costs.
Solution: Analysis of the reports, roadmap and management model; a new architecture, migration of the developments to a lakehouse in Fabric with integrity validation, auditing and load control, row- and column-level security for sensitive data, and data governance implemented on Azure with Purview.
Result: Centralised reports, lower licensing and development costs, an analytics layer for self-service and data services for third parties, and a foundation ready for machine learning.

Microsoft Fabric
Power BI
Purview
SAP R/3
SAP BusinessObjects
PySpark

Data platforms

International door-to-door logistics operator

AI-powered digital platform on top of the ERP

Challenge: More than 30 price requests a day by email and phone, rates requested by hand from five or six suppliers with up to two days' wait, consolidation in Excel and no shipment visibility for the client.
Solution: External quoting and booking portal, internal rates portal, automated quoting engine and control tower on top of the ERP, with predictive analytics and data governance.
Result: Clients quote, book and track their shipments without calling, and rates go from two days of emails to a real-time comparison panel. A 24-month programme in two-week sprints.

Odoo
Python
Predictive analytics

Odoo and business digitalisation

Real-time digital advertising company

Migration to a cloud-native architecture with Databricks

Challenge: The data and analytics platform was not keeping pace with business growth: integration and reporting processes needed modernising, and the teams had no experience with the new technology.
Solution: Target architecture on AWS, migration of the integration and reporting processes to Databricks, and knowledge transfer to the teams.
Result: Scalable storage with orchestrated pipelines, a foundation ready for machine learning, and the client's teams running operations autonomously.

AWS
Databricks
Python

Data platforms

HR department of a city council

AI assistant for Human Resources in Microsoft Teams

Challenge: More than 60 requests a week from staff about sick leave, absences, holidays and the collective agreement. Every query and every supporting document was read, interpreted and classified by hand.
Solution: A conversational assistant in Teams with RAG over the collective agreement and three years of real requests, which processes attached documents and asks HR for approval when in doubt.
Result: Immediate answers for employees with the source cited, less repetitive work for HR, and human decisions where they are needed. A foundation that can be extended to other internal procedures.

Microsoft Teams
RAG

AI automation

Large local authority with its own data office

Selecting the data and AI governance tool

Challenge: From among the leading data governance tools on the market, the one that best fitted the client had to be selected to govern BigQuery and the dashboards, and to integrate the data governance tools available in the cloud.
Solution: Maturity assessment, requirements catalogue and weighted scoring matrix; ten tools evaluated with the vendors and cross-checked with Microsoft, Google and Anjana.
Result: Three finalists and a reasoned decision, reusable criteria, and compliance assessed against UNE 0077-0080, UNE 0085 and the EU AI Act.

Microsoft Purview
BigQuery
Power BI
Fabric

Data governance

Several ministries of Spain's central government

Common criteria for choosing a data governance tool

Challenge: Data governance initiatives running in parallel across different ministries, with no common criteria for choosing a tool and the requirement to fit public administration regulations.
Solution: Gathering of functional and technical requirements for each ministry, comparison of the tools on the market, open-source ones included, assessment of data maturity against the UNE standards and a recommendation for each usage scenario. Theoretical and hands-on training on real cases, and support during roll-out.
Result: A decision backed by consistent criteria reusable across ministries, and a basis for starting implementation without redoing the analysis.

Requirements
Market comparison
Open source
UNE 0077-0079

Data governance

Public body managing data from more than 200 organisations

Purview and automatic documentation with AI

Challenge: Fabric, SQL Server, AWS S3 and Purview coexisting, unmonitored loads, no guaranteed quality, and knowledge concentrated in suppliers that was lost with every turnover.
Solution: Maturity audit and target organisational model, phased roadmap, Purview with domains, data products and glossary, and AI accelerators to document the models.
Result: Documentation generated automatically with minimal supervision, owners and roles identified, a two-year roadmap and a self-sufficient technical team.

Microsoft Purview
Fabric
SQL Server
AWS S3

Data governance

Bank

Data governance and information architecture

Challenge: More than 60,000 tables neither classified nor organised by subject area, with no catalogue accessible to the business and no common modelling criteria.
Solution: Classification by subject area with erwin Data Modeler, a catalogue for business and technical users, an architecture review and a methodological framework based on DAMA and Kimball.
Result: A governance methodology applicable to every platform, data the business can understand, and modelling defects identified.

erwin Data Modeler
Teradata
DAMA
Kimball

Data governance

Multinational executive search and recruitment firm

Operations dashboard connected to the database

Challenge: The dashboard read from manually generated Excel files, with no history to compare KPIs and a model full of unused fields and relationships.
Solution: A gateway on AWS to connect Power BI directly to the database, a redesigned model, three years of history and global filters by sector, business unit and client.
Result: Up-to-date data without manual files, historical comparisons and consistent KPIs, and users trained to make the most of it.

Power BI
AWS

Data and analytics

Road passenger transport operator

AI transformation roadmap

Challenge: A growth plan with technology as an ally, and the need to know which processes to improve and what AI could automate across the company.
Solution: Review of processes in Operations, IT, Sales, Finance and HR through workshops and sessions with management, and a technology recommendation for each challenge: generative AI, RAG, agents and hyperautomation.
Result: Specific reports and roadmaps for each area, with the AI use cases prioritised.

Generative AI
RAG
Agents

Digital strategy

Specialist veterinary clinic

Digitalisation plan for clinical and administrative processes

Challenge: Processes with a lot of manual handling, clinical and financial information scattered across tools, and no management indicators.
Solution: Automation opportunities, document digitalisation, integration of clinical and financial information, target processes and management indicators.
Result: End-to-end traceability, less administrative work, and decisions based on data rather than impressions.

Automation
Dashboards

Digital strategy

Public university

Governance portal for a federated data space

Challenge: The use cases depended on proprietary GIS technology, creating barriers to entry for new municipalities, and the chain of trust and the catalogue had to be managed autonomously.
Solution: A neutral governance portal, software for the provider, promoter and consumer roles, a federated catalogue on EDC and the DSP protocol, and GAIA-X credentials.
Result: New municipalities join without GIS licences, and the governance logic lives in the data space itself.

EDC connectors
DSP protocol
GAIA-X
Python
PostgreSQL
Docker

Data platforms

Financial advisory firm

Automated email and document processing with AI

Challenge: Hours every day sorting email and keying in invoice and case file data.
Solution: Classification and routing of incoming email, document extraction and case file matching with AI.
Result: The team reviews exceptions instead of typing data.

Document extraction
n8n

AI automation

Professional services consultancy

Project profitability and working-time recording with TIME

Challenge: Hours were logged in spreadsheets and clocking-in lived in a separate system: no visibility of which projects and clients made money, no reconciliation between clocked and logged hours, and the legally required working-time record handled by hand.
Solution: Roll-out of TIME: daily time logging per project, date-based rates and costs, margin and utilisation reports by project, client and person, and electronic recording of each employee's working day (clock-in, clock-out and breaks) with the official work calendar and a monthly report ready for a labour inspection.
Result: Staffing and pricing decisions based on real margin, clocked and logged hours in one place, and a digital working-time record ready to adapt to Spain's upcoming working-time recording rules.

TIME
Time recording
Margin reports

TIME

Regional cybersecurity agency

A data governance model with a first data domain and product

Challenge: Many data sources at different levels of maturity and an architecture not designed to govern them: no shared view of what data existed, who was accountable for it and how to use it with confidence.
Solution: Assessment of the maturity of the data sources and of the existing architecture, gap analysis, design of the architecture supporting data governance, and roll-out of the model with a first data domain and product.
Result: A data governance model in place, prioritised gaps and a working data product that serves as the pattern for the next domains.

Data maturity
Data architecture
Data domains and products

Data governance

Spanish central government ministry

Data governance training and OpenMetadata roll-out

Challenge: Technical and business teams had to take on data governance without a shared set of concepts or a catalogue tool in place.
Solution: Training sessions for technical and business teams on data governance methodology and techniques, and roll-out of OpenMetadata as the catalogue.
Result: Teams sharing the same data governance language and criteria, and a working open-source catalogue to build on.

OpenMetadata
Training

Data governance

Diversified business holding

From SAP BusinessObjects to Power BI on Microsoft Fabric

Challenge: The group's dashboards lived in SAP BusinessObjects on top of SAP R/3: expensive to maintain, slow to evolve and hard to combine with the rest of the group's data.
Solution: Migration of the reporting layer to the Microsoft Fabric ecosystem: load notebooks in PySpark and new Power BI dashboards, with semantic models and row- and column-level security.
Result: Power BI dashboards on the group's lakehouse, a single version of the data and a report catalogue that is easier to maintain and extend.

Power BI
Microsoft Fabric
PySpark
SAP BusinessObjects
SAP R/3

Data and analytics

Shall we talk about your project?

A first 30-minute conversation, no strings attached.

Tell us about your project WhatsApp