Insight

Reimagined: What if we rebuilt our best work with today’s tools?

Over the next two months, we’re doing something a bit different. We’re looking back at our project catalogue and asking: how would we build this today?

The world is moving at a breakneck speed. As such the tools and means to create those projects have changed. AI, natural language processing, real-time data pipelines. What took months now takes weeks. What was a prototype, can now be a reality.

So we’re reimagining four projects from our back catalogue. Each one tackles a real problem we’ve seen across multiple sectors. Each one shows what ‘data intelligence, beautifully designed’ actually means when you apply it to messy, real-world challenges.
This isn’t about showcasing technology for its own sake. It’s about showing what becomes possible when you combine the right tools with clear thinking about what users actually need.

Urban Ocean Lab: Turning a document repository into a knowledge engine

In 2023, we collaborated with Urban Ocean Lab to create a resources hub for coastal city policymakers. The platform surfaces hundreds of materials; datasets, case studies, research papers, to inform climate and ocean policy.
The challenge wasn’t the content. Urban Ocean Lab had done the hard work of curating rigorous, practical research. The challenge was discovery. How do you help a policymaker in Miami find the exact insight they need from a repository of 400+ documents?
The original solution was solid: clear categorisation, sensible tags, clean search. But it relied on users knowing what they were looking for. And most users don’t.

The problem isn’t unique

We’ve seen this pattern across sectors. The Partnership Against Child Exploitation faces the same challenge with humanitarian protection research. Large pharmaceutical companies struggle with internal knowledge bases. Government departments sit on decades of policy documents that nobody can efficiently navigate.

The pain points are consistent: vast repositories with minimal content preview beyond titles. Simple string-based search that misses context and meaning. Users redirected to external resources rather than receiving synthesised insights. Hours of manual research to answer questions that should take minutes.

The reimagination

What if the corpus itself became navigable? Not just searchable but explorable. Here’s how we would reimagine it today:

1. Visual corpus intelligence. When a user lands on the platform, they see the landscape of knowledge at a glance. Topic clusters show which research areas dominate. Geographic coverage maps reveal where evidence is strongest (and where gaps exist). A timeline shows how focus areas have shifted over time. Before searching for anything, users understand what’s there to find.
2. Intelligent query interface. Move beyond string matching. A natural language interface that understands context: “What approaches have worked for flood resilience in Southeast Asian cities with populations under 2 million?” The system doesn’t just find documents, it synthesises insights across them, with clear attribution so users can verify and dig deeper.
3. Smart content discovery. Help users find what they didn’t know they needed. Dynamic feeds surfacing documents relevant to current news. Related content suggestions based on what users are actually exploring. Quick insight cards that provide key findings before users commit to reading a full report.

How this works

The technical approach combines proven AI tools with modern data visualisation in a straightforward way. We process the document repository using available AI services to extract key themes and concepts, then cluster documents by topic rather than relying solely on manual tags. This creates a visual map where users can see topic clusters as interactive bubbles, sized by the number of documents, coloured by theme, and clickable to filter results.

The intelligent search layer enhances traditional keyword matching with contextual understanding. When a user asks “What approaches have worked for flood resilience in Southeast Asian cities?”, the system uses AI APIs to understand the query context, retrieves relevant documents, and provides a synthesised response with clear source attribution. It’s not custom AI development, it’s smart application of familiar tools like OpenAI or Claude, integrated with clean visualization and intuitive interface design.

The innovation lies not in inventing new technology, but in combining existing tools thoughtfully to create something that feels both sophisticated and immediately useful. It’s the difference between a document database and a knowledge ecosystem, achieved through better, more thoughtful design, not revolutionary engineering.

Why this matters

The gap between ‘organised’ and ‘intelligent’ is where most knowledge systems fail. Tagging and categorising is necessary but not sufficient. True intelligence means understanding relationships between concepts, anticipating user needs, and presenting information in ways that accelerate understanding rather than creating work.

This isn’t about replacing human expertise. It’s about removing the friction that prevents experts from applying their expertise. A policymaker’s job isn’t to excavate research repositories—it’s to make good decisions, and every hour spent searching is an hour not spent thinking.

What comes next

Over the coming weeks, we’ll share three more reimagined projects:

CyberSeek — What happens when static job market data becomes personalised career intelligence?

Sport England — Can real-time participation data transform how we measure policy impact?

Nest — How do you turn raw energy data into visual narratives that actually change behaviour?

Each tackles a different sector, a different data challenge, a different set of users. But the through-line is the same: taking what we know about data visualisation and adding the intelligence layer that modern tools make possible.

Data intelligence, beautifully designed. That’s what we’re building toward.