References
Explore the Connections
Answers You Can Trust Transition Digital Brazil Prototype App is built for decision-makers who can’t risk guesswork. Every answer is drawn from trusted sources and shown with evidence you can verify.
How Transition Digital Works
Transition Digital is built for decision-makers who can’t afford guesswork. Every response is grounded in curated data and comes with evidence you can verify. Behind the scenes, the system follows a clear path:
- Understand your question. Your query is first analyzed to understand what kind of information you need and which datasets are most relevant.
- Retrieve the evidence. The system uses retrieval techniques and knowledge graphs to find the exact passages, records, or data points that matter.
- Assemble the answer. Results are woven together into a coherent answer, complete with inline citations and—when helpful—visuals like maps, charts, or tables.
This isn’t just about producing an answer; it’s about showing a transparent and auditable the trail of evidence, so you always know where the information comes from, and trace how every insight was produced.
Sources
For this pilot, we have focused on a single geography — Brazil. The prototype draws on a curated set of the most relevant and trusted datasets across science, policy, corporate reporting, and modelling domains. Together, they provide a multidimensional view of climate and development dynamics.
Science and environmental data
- Science Panel for the Amazon (2021) — landmark peer-reviewed regional report.
- PRODES Deforestation — official annual deforestation rates.
- ERA5-Land Heat Indices — recent surface temperature and heat stress records.
- WMO State of the Climate in Latin America and the Caribbean 2024 — climate indicators and regional risks.
- MapBiomas Deforestation Report 2024 — high-resolution deforestation alerts and land-use change data.
Policy and governance data
- Climate Policy Radar Knowledge Graph — searchable national laws, policies, and UNFCCC submissions such as NDCs.
- NDC Align — comparison of Brazil’s national climate ambition against governance and implementation capacity on the national and state level.
Corporate and economic data
- GIST Environmental Impact — biodiversity and emissions metrics by company.
Modelling and systems data
- TransitionZero Solar Asset Mapper — facility-level solar deployment data.
- IPCC AR6 (selected chapters) — scientific markers and scenarios for Latin America and Brazil.
These datasets form the initial foundation of TD’s shared data layer, which will expand through ongoing partner contributions and integrations — enabling richer, cross-domain insights over time.
What this prototype Transition Digital will never do
- Hide how an answer was created
- Provide results without citations or context
With Transition Digital, the goal isn’t just to answer your question—it’s to give you confidence that the answer is right, relevant, and verifiable.
Looking Ahead
Each stage of this prototype hints at where we're headed: a shared, transparent infrastructure for climate and development intelligence — one that grows through collaboration and continual improvement.
We're already working with a dozen trusted datasets, and are constantly expanding the range and depth of sources across regions and topics. Over time, these will form a living network of shared datasets contributed by TD partners and external collaborators, creating a richer and more diverse foundation for insight generation.
Our knowledge graph, developed by Climate Policy Radar, connects related concepts, policies, and data so the system can understand meaning and context rather than relying on keywords alone. As partners add new taxonomies and domains, this graph will make it easier to connect insights across sectors and geographies.
Learn more about how the knowledge graph works →
We're also building toward greater openness and participation — introducing a self-serve layer that will allow builders to plug in new data and create specialised applications, all while maintaining transparency and quality. See all our datasets →
Future versions of TD will handle more data types and modalities, apply deeper quality checks and version control, and refine the protocols that determine how questions are answered — always preserving what makes this system trustworthy: transparency, evidence-based synthesis, and explainability. To learn more about Transition Digital and how to get involved, visit Transition Digital
Methodology & Datasets
System Architecture
Transition Digital Brazil Prototype App combines multiple data sources through a Model Context Protocol (MCP) server architecture. The system uses AI-powered Retrieval Augmented Generation (RAG) to intelligently query and synthesize information from diverse climate-related datasets. We maintain consistent references and citations.
How It Works
- Query Processing: User queries are processed by an AI model that determines which data sources and tools are most relevant.
- Dynamic Tool Selection: Our MCP server provides access to specialized tools for each dataset, including:\n * Knowledge graph queries for policy documents
- Geospatial queries for facility locations
- Statistical analysis for trends and comparisons
- RAG Integration: The system uses semantic search and knowledge graph traversal to find relevant information, then augments responses with data from multiple sources.
- Response Synthesis: Results from different tools are combined into a coherent response with appropriate visualizations (charts, maps, tables) and inline citations.
Data Sources
Transition Digital Brazil Prototype App integrates curated climate, nature and development data from authoritative sources. Data includes: climate laws and policies and climate governance data, environmental impact data, renewable energy infrastructure, deforestation monitoring, climate indicators, and authoritative scientific reports.
This app features datasets relevant mainly to Brazil.
For detailed information about our data sources, methodology, and update schedules, please visit our Technical Methodology Documentation.