A new chapter for Uwazi

Something big is happening with Uwazi.

Over the past several years, we have helped Uwazi grow from a tool developed alongside human rights organisations into infrastructure used by hundreds of organisations to document violations, manage cases and complaints, investigate public interest information, preserve evidence, build archives, and make complex collections of information accessible.

And we want many more organisations to be able to use it.

But scaling Uwazi is not simply a matter of adding more servers or signing up more users. It is one of the reasons why we have undertaken a much deeper revamp that will become Uwazi 2.0. After years of development and feedback from those who use Uwazi every day, we understand better than ever what they need, and where the technology itself needs to evolve. We are bringing those lessons into everything: from the user experience to the underlying application architecture. The result will be a faster, more intuitive and more powerful Uwazi, built on foundations designed for what comes next.

So that is what we have been doing.

Building from the foundations up

A lot of the most important work happening in Uwazi right now is not immediately visible at first. We have been rebuilding and modernising significant parts of the underlying technology, strengthening the codebase and rethinking some of the core components that have been powering Uwazi for years. That work is creating the foundation for a new generation of Uwazi.

Some of the changes are already implemented. Some of it is coming soon.

The new Library
The Library is the heart of an Uwazi collection: the place where users find, browse, filter and work with their documents and data. We are redesigning the Library around clearer visual patterns and more intuitive workflows, alongside a major accessibility overhaul with a more responsive interface, dark mode and themes. Full-text search results are also being redesigned to make large collections easier to explore and understand. Looking ahead, we are working towards timeline visualisations and time-based filtering, giving users new ways to see patterns across their information.

Rebuilding the entity experience 
In Uwazi, an “entity” is simply a record: it might be a person, an organisation, a court case, an incident, a recommendation, a piece of legislation or a document. We are rebuilding the Entity experience from the ground up, with clearer visual patterns and accessible, responsive design in line with the latest standards. We are also rethinking how users manage attached files and documents, including better ways to navigate tables of contents and work with text references. New relationship fields will make connections between records clearer and more meaningful, including grouping related information by relationship type. The result is a more intuitive way to work with complex information while retaining the flexibility that has always been central to Uwazi.

Connections and relationships
Human rights information rarely exists in isolation. A person may be connected to an incident; an incident to a location; a court decision to another case; a recommendation to the state responsible for implementing it. Relationships allow Uwazi to represent those connections. We are rebuilding the underlying technology using PostgreSQL, ParadeDB and Apache AGE, giving Uwazi true graph-querying capabilities. This will allow users to move beyond exploring one connection at a time and follow relationships across multiple steps. For example, users will be able to better trace how a person, event, organisation, location and piece of evidence are connected. Over time, we plan to support shortest-path and hierarchical exploration as well. These capabilities will be matched by a redesigned interface that makes complex networks of information something researchers can actually see, navigate and use.

Smoother text referencing
Sometimes the important connection is not between two documents, it is between particular passages within them. Text references allow users to connect specific pieces of text to other information in their collection, making it possible to trace sources, supporting evidence and related material much more precisely. We have introduced a new visualisation system that also shows where those references appear within a document, including which pages contain them and where references are most concentrated. We have also redesigned how references are created, with a clearer and smoother workflow for selecting passages and making connections.  

And all of this is arriving alongside a fresh Uwazi design, making the platform cleaner and easier to navigate.

Making Uwazi easier to access

We are also changing how organisations get Uwazi in the first place.

Our goal is to make it possible for many more civil society organisations to use Uwazi as part of their documentation and information infrastructure, without needing a lengthy technical setup process.

That means we will be moving towards automatic instance creation directly from the Uwazi website, alongside automated subscription and billing. This is currently a work in progress and we will keep our existing partners informed as we move forward.

In much plainer language: we want an organisation that needs Uwazi to be able to choose  to create an instance and get started.

And getting started should actually be easy.

That is why we have also launched a new Uwazi documentation portal, and we are continuing to rethink how we document the platform itself. We want guidance that is clear, practical and easy to navigate. Whether you are setting up your first collection, designing templates, building relationships between information or trying to understand a new feature, you can find the answers you need directly in Uwazi Docs.

Ultimately, scaling Uwazi is not just about making the technology available to more organisations. It is about enabling organisations to use it independently and confidently.

Making powerful technology genuinely useful

While we are strengthening the foundation and accessibility of the tool, we  we are continuing to build machine learning features into Uwazi; not because AI is something every platform suddenly needs, but because there are very real documentation problems where these technologies can save enormous amounts of human time and make large collections of information more usable. This is about choosing the right tool for the right job. 

The next stage is making these capabilities available to more Uwazi partners.

That includes:

Metadata extraction
Instead of reading hundreds or thousands of documents and manually copying information into database fields, Uwazi can help identify and extract information such as names, dates, organisations, locations or other specific information, for human review.

Paragraph extraction
Sometimes you don't need an entire document,  you only need the sections that answer a particular question. Extraction tools can help identify relevant passages across large numbers of documents, dramatically reducing the amount of material researchers have to review manually.

Named entity recognition
Uwazi can identify information such as people, organisations and locations mentioned within documents, helping turn large volumes of unstructured text into information that can be organised and explored.

Automatic translation
Collections often span multiple languages. Machine translation can help users quickly understand and work across languages, making information more accessible to researchers and communities who would otherwise struggle to use it.

Semantic search
Traditional search is very good at finding the words you typed. Semantic search can help find information based on what those words mean. That means a researcher can potentially discover relevant material even when a document uses different terminology from their search.

Bert and text classification
Bert is our very own integrated agent, named after Bert Vertstappen - our long time documentalist and now Board member. Our digital Bert is beginning to do more than answer questions. Today, it can assist users with administrative tasks inside Uwazi, such as creating or modifying entities and documents, managing users and pages, and working with data visualisations. It can also answer more complex questions about the information held in a collection, drawing on entity properties and library statistics. Bert’s capabilities are still developing. For example, it cannot yet read the documents themselves, but as we progressively open more Uwazi tools to Bert, it will be able to assist with an expanding range of tasks. The goal is an agent that works within Uwazi alongside the researcher, helping them navigate, understand and manage their information while keeping the human firmly in control.

Document conversion and PDF segmentation
PDFs are ubiquitous in human rights work and notoriously difficult to work with as data. We are developing better ways to understand their structure and convert their contents into accessible, structured formats such as HTML, recognising text, images, tables and sections along the way.

These are not features designed to replace researchers, documenters or analysts. They are designed to help people navigate overwhelming quantities of information and spend more of their time on the work that requires human expertise and judgement.

Why are we doing all of this?

We believe that documentation infrastructure matters.

Civil society organisations are collecting extraordinary amounts of information: testimony, evidence, court decisions, recommendations, legislation, reports, complaints, investigations, images, videos and more.

The challenge is no longer simply how to store it. It is how to structure it. Connect it. Search it. Understand it. Preserve it. Share it safely. And ultimately, use information for purpose.

We want Uwazi to be infrastructure that civil society can rely on to do that: open source, flexible, rights-aligned, and capable of growing alongside the organisations that shape and use it.

There is a lot still to come. We can't wait to show you what we've been building.

Watch this space.