When AI Becomes Reality: From Need to Working Prototype in Weeks – Together with World Childhood Foundation

Client: World Childhood Foundation
Industry: Non-profit
Solutions Area: AI

A Knowledge-Rich Organisation with Many Documents

World Childhood Foundation is a nonprofit foundation that works to prevent sexual abuse against children through long-term projects and partnerships in Sweden and internationally. The organisation runs and monitors numerous initiatives with local partners across multiple geographical and thematic areas.

That also means a large amount of valuable information: documents, reports and historical material accumulated over time. The challenge was not a lack of data — but that this material was scattered across multiple systems and file repositories. Quickly gaining an overview, finding relevant history or generating decision- and report-ready summaries required significant manual effort.

Childhood was already familiar with the potential of AI and had explored it in strategic contexts as early as 2019, helping lay the foundation for Stella Polaris, an initiative aimed at coordinating and accelerating AI efforts to strengthen the fight against sexual abuse. In this project, the question was not whether AI was relevant, but how the technology could be used responsibly, transparently and concretely to support and improve internal processes.

The Goal of the Engagement

The assignment was designed to:

• investigate whether AI can create practical value from existing data
• run a small-scale test using real questions and real information
• produce a basis for decision-making about future investments — rather than deliver a finished product immediately

From Early Strategic Interest to Internal Impact

The challenge was both technical and methodological. The data existed – but it was difficult to use in a consolidated way. Much time was spent searching and reading rather than analysing, drawing conclusions and making decisions.

AI was seen as promising but abstract. The key questions were: What does this look like in practice? How do you get started? And how do you know if it’s worth investing more?

The need was to test whether existing data could create value with the help of AI – in a way that was accessible and experience-based rather than theoretical.

A Proof of Value – Not a Large Commitment

HHiQ proposed a Proof of Value approach that emphasised rapid learning, clear scope and collaborative exploration rather than comprehensive implementation.

The work began by jointly clarifying needs, goals and criteria for success. Relevant use cases were identified, and existing documents were prepared as the data basis.

From there, we built a simple but tailored AI prototype combining document search with AI-generated responses. The protoype was deliberately scoped and designed to be reliable and transparent: it

• answered only based on Childhood’s own documents
• always showed sources and references
• stuck to facts without hallucinating or inventing content

A key part of the process was demonstrating and testing the prototype together. Participants were able to ask the kinds of questions they encounter in their everyday work, explore opportunities and limitations, and experience the technology in realistic scenarios.

Deliverables and Next Steps

In addition to the prototype, HiQ delivered recommendations for future technical, data and organisational choices – providing a clear basis for how the AI initiative could progress.

Technology That Enables Without Taking Over

Technical choices were based on combining proven tools with modern AI capability, without building more complexity than necessary. The solution was built using cloud-based services that can be scaled up if needed:

• AI engine: Claude AI for document-based responses
• Backend: FastAPI in Python
• User interface: Streamlit for a simple and interactive experience
• Infrastructure: Docker and Azure Container Apps

Work was carried out iteratively over roughly four weeks — from needs identification and document preparation to development and joint testing with real questions. The result was a quick yet stable prototype, with strong potential for further development.

AI That Becomes Tangible

The project resulted in a functioning AI prototype that showed how existing information can be made more accessible, searchable and usable in everyday work. Equally important was the shared understanding that arose around what AI can actually contribute – and where its limitations lie.
By starting with real questions and real data, AI became something experienced together rather than an abstract side project. The threshold for adoption was lowered, engagement increased and the technology led to a concrete tool and foundation for discussion.

Key Learnings

The prototype worked particularly well when summarising and responding to fact-based questions tied to specific projects, time periods or themes. Users found the interface intuitive and easy to adopt.

As Childhood themselves stated:

As Childhood themselves stated:

“Although it was a prototype, we were able to test the application online and work with a solution adapted to our needs.”

The collaboration also generated ideas for future steps — including project and status overviews based on metadata, more role-tailored responses, support for visualisations such as charts and tables, and advanced data engineering for higher data quality.

A First Step With a Forward Look

The project showed that even a scoped prototype can create clear value, build knowledge and serve as a starting point for a long-term AI journey.

Nyfiken på hur ni kan testa AI i liten skala – och skapa verkligt värde snabbt? Låt oss ta ett samtal!