Pulp produced in Brazil in 2024
TOMI AI FORESTRY · IN DEVELOPMENT
Building domain intelligence for Brazilian forestry.
TOMI AI Forestry is Amazonia Tech Partners' domain-specific AI project for silviculture and forest operations. It brings together a governed technical-scientific knowledge base, retrieval-augmented generation, specialist evaluation and a controlled path from research to real-world deployment.
Planned market availability: Q1 2027 through KENE — The Sovereign AI Factory, beginning with the Brazilian forestry sector and subject to development and validation milestones.
AI built for one of the world's most consequential forestry economies.
Why this mattersBrazil produced 25.5 million tonnes of pulp in 2024, ranking as the world's second-largest producer and leading global exporter. Maintaining that position increasingly depends on the ability to transform scientific and operational knowledge into systems that can support people, organisations and machines with context.
TOMI AI Forestry is being built for that challenge: not as a generic model with forestry vocabulary, but as domain intelligence grounded in the scientific, technical and regional realities of Brazilian silviculture.
Brazil's position in pulp production
Brazil's position in global pulp exports
A plausible answer is not necessarily a forestry-grade answer.
Context changes the answerForestry knowledge changes with the physical context.
Silvicultural decisions depend on species, genetic material, site conditions, climate, region, management regime, equipment and operational constraints. Recommendations that are valid in one context can be incomplete or unsafe in another.
That is why TOMI AI Forestry is designed around more than text generation. The development process separates source quality from answer quality, traces responses back to evidence and places specialist judgement inside the model-improvement loop.
System principle
Built for forestry. Tested against evidence. Designed for the real world.
A structured knowledge base for forestry intelligence.
Delivered foundationThe project's first macro-delivery has been completed: a structured technical-scientific knowledge base organised through a hierarchical map of domains, themes and subthemes.
The current architecture spans silviculture, forest harvesting and logistics, with branches covering implementation, nurseries, forest protection, forest improvement, roads, yards, transport, safety, maintenance and systems. Scientific publications and technical materials are classified to support governed retrieval, evaluation and continuous expansion.
Knowledge mapped from domains to specific questions
Knowledge organised from forestry domains to operational themes and specific questions.
Scientific and technical sources under classification
Scientific and technical sources classified by subject, period, language and access conditions.
Answers remain connected to supporting passages
A working retrieval environment connects generated answers to documents and supporting passages.
SilvicultureForest harvestingLogisticsImplementationNurseriesForest protectionForest improvementRoadsTransportSafetyMaintenanceSystems
From curated sources to specialist-validated reference data.
Evidence before confidenceThe current development loop is designed to reveal not only whether an answer is wrong, but why it is wrong and what must improve.
Curate
Organise scientific and technical sources by domain, theme and subtheme.
Retrieve
Use retrieval-augmented generation and vector search to identify relevant evidence.
Answer
Generate a response with the documents and passages used by the system.
Evaluate
Ask specialists to assess correctness, completeness, context and source sufficiency.
Correct
Record the technical explanation, reference answer and most appropriate supporting source.
Improve
Use the reviewed data to build benchmarks, diagnose failure modes and guide later model adjustment.
Curated sources→RAG retrieval→Answer + sources→Specialist review→Reference data
Development boundary
TOMI AI Forestry is being developed through governed knowledge, specialist evaluation and controlled delivery. Public performance claims will follow the relevant validation stages.
Different capabilities, organised around one development path.
Science · Engineering · Sector connectionAmazonia Tech Partners
AI engineering, product architecture, controlled infrastructure and the path from technical validation to market delivery.
UFV–DEF and UFV–Fibras Florestais
A forestry research environment connected to the Department of Forestry Engineering at the Federal University of Viçosa and its accredited EMBRAPII Unit.
SIF
A university–industry interface with more than 30 associates across the forest sector, supporting the connection between research, innovation and sector needs.
EMBRAPII / Sebrae mechanisms
The project's R&D structure references EMBRAPII/Sebrae innovation mechanisms alongside private execution and investment.
Institutional boundary
Institutional participation supports research, development and technical-scientific work. It does not constitute certification, commercial endorsement or a guarantee of model results.
Help build the reference layer for forestry AI.
Companies and researchersTOMI AI Forestry is entering the stage where high-quality questions, contextual expertise and rigorous evaluation can materially improve the system. ATP is opening conversations with companies and researchers interested in contributing to the project's technical development.
- propose real technical and operational questions;
- contribute public or explicitly authorised technical material;
- help prioritise relevant use cases;
- discuss future technical validation and pilot participation.
- contribute scientific literature and domain expertise;
- identify regional, biological and operational context that generic models miss;
- participate in question design, specialist review and benchmark development;
- help define the evidence required for technically responsible answers.
Participation boundary
Discuss a contribution →
The current collaboration focus is technical-scientific knowledge, not confidential production data. Contributions, access and pilot participation are subject to project governance, rights clearance and agreed conditions.
Domain intelligence on controlled infrastructure.
Sovereign delivery pathKENE — The Sovereign AI Factory
Planned controlled delivery for domain-specific AI.
TOMI AI Forestry is planned for delivery through KENE — The Sovereign AI Factory, ATP's controlled infrastructure layer for domain-specific AI. The objective is to support deployment choices that match the security, governance and operational requirements of the Brazilian forestry sector.
Planned market availability Q1 2027. Timing and access conditions remain subject to technical validation, infrastructure readiness and project governance.
A separate next initiative for native-forest intelligence.
TOMI AI Native ForestsConsortium initiative under structuring
TOMI AI Native Forests
TOMI AI Forestry focuses on silviculture and planted-forest operations. TOMI AI Native Forests is a distinct next initiative intended to address the different scientific, ecological and governance requirements of native forests.
Its consortium, team and funding structure are currently being organised, with work planned to begin in Q4 2026. This initiative is not part of the current TOMI AI Forestry delivery scope.
Discuss research collaboration →From knowledge to domain intelligence
Build forestry AI against evidence, not assumptions.
If your organisation can contribute technical knowledge, specialist evaluation or a relevant validation environment, talk to Amazonia Tech Partners about TOMI AI Forestry.