01THE RESEARCH POSITIONDOMAIN BEFORE MODEL
01 / The research position
The domain comes before the model.
Climate, forestry, industrial resilience and sovereign systems cannot be reduced to a general-purpose model connected to a document library. They require domain methods, governed evidence, specialist judgement and infrastructure designed around the operating environment.
ATP begins with the problem as it exists in the physical economy. We structure the knowledge, data, rules and evaluation criteria that govern the domain. Models and agents are then developed inside that architecture and connected to the applications and infrastructure required for controlled delivery.
Research begins with the evidence, methods and constraints that govern the domain.
Domain framing
Define the operational problem, decision context, evidence requirements and institutional boundaries before selecting a model architecture.
Knowledge architecture
Organise specialist knowledge through taxonomies, ontologies, source hierarchies and explicit relationships between methods, evidence and decisions.
Governed intelligence
Build retrieval, deterministic services, models and agents around controlled sources, documented evaluation and clear responsibility boundaries.
Controlled delivery
Connect specialised intelligence to applications, private environments and region-aware infrastructure according to the requirements of each engagement.
02PROGRAMME REGISTRYMATURITY IS EXPLICIT
02 / Programme registry
One research system. Different maturity states.
ATP separates systems available for use from active R&D, programmes under structuring, consortium initiatives and strategic infrastructure. Maturity is stated explicitly because technological direction is not the same as commercial availability.
R.01FORESTRY AND SILVICULTUREIN DEVELOPMENT
TOMI AI Forestry
Domain intelligence for the Brazilian forestry sector.
TOMI AI Forestry is being developed to organise and apply specialist forestry knowledge across silviculture, forest operations and technical decision support. It demonstrates ATP’s method for turning sector knowledge into a governed intelligence environment.
- PUBLIC EVIDENCE
- The current foundation includes a hierarchical forestry knowledge architecture, an initial corpus macro-delivery, a working retrieval environment and an initial answer-and-source set entering specialist validation.
- BOUNDARY
- Planned market availability is Q1 2027 through KENE — The Sovereign AI Factory, beginning with the Brazilian forestry sector and subject to development and validation milestones.
R.02CONTROLLED AI INFRASTRUCTURESTRATEGIC INFRASTRUCTURE
KENE — The Sovereign AI Factory
The infrastructure layer for specialised and region-aware AI delivery.
KENE is ATP’s strategic infrastructure direction for developing, evaluating and serving domain-specific models in controlled environments. It connects model development, private deployment requirements and the operational boundaries of institutions working with sensitive knowledge and data.
- PUBLIC EVIDENCE
- KENE provides the infrastructure direction through which ATP intends to move specialised models from governed R&D into controlled delivery environments.
- BOUNDARY
- Capacity, commercial availability, data-residency configuration and region-specific deployment remain engagement-specific and are not represented here as universally available.
R.03GREEN INDUSTRY, CLIMATE RESILIENCE AND SUSTAINABLE FINANCEPROGRAM UNDER STRUCTURING
GRID — Green Resilience & Industrial Decarbonization
A programme architecture for industrial climate transition.
GRID is being structured to connect industrial diagnostics, emissions management, climate risk, applied R&D, transition planning, project preparation and sustainable finance. Its initial programme direction focuses on industrial resilience and decarbonisation in Northern Brazil.
- PUBLIC EVIDENCE
- A working programme architecture defines the intended pathway from mobilisation and diagnosis to applied research, implementation readiness, finance preparation and measurement.
- BOUNDARY
- The pilot geography, participating organisations, financing structure, budget, implementation targets and partner roles remain subject to confirmation and formal programme design.
R.04MEDITERRANEAN AGRICULTURE AND CLIMATE ADAPTATIONCONSORTIUM INITIATIVE UNDER STRUCTURING
OLEA AI
A proposed domain-intelligence programme for climate resilience in the European olive sector.
OLEA AI applies ATP’s domain-intelligence method to the knowledge, agronomy and climate risks of olive cultivation. The proposed programme combines a governed knowledge environment, specialised AI agents and future integration with field and environmental data.
- PUBLIC EVIDENCE
- The initiative has a defined concept, technical proposal and research pathway for adapting ATP’s vertical-AI method to the European olive sector.
- BOUNDARY
- OLEA AI is not a commercially available product or a funded project. Consortium participation, funding, pilot sites, technical delivery and expected outcomes remain subject to formal confirmation.
R.05NATIVE FORESTS, BIODIVERSITY AND ENVIRONMENTAL INTELLIGENCECONSORTIUM INITIATIVE UNDER STRUCTURING
TOMI AI Native Forests
A separate research direction for intelligence in native forest systems.
TOMI AI Native Forests is the planned follow-on initiative to develop governed intelligence for native forests, biodiversity, biomass, carbon and environmental observation. It is separate from the current scope and development state of TOMI AI Forestry.
- PUBLIC EVIDENCE
- ATP has defined the initiative’s domain and intended role within its nature-intelligence direction. Consortium, team and funding arrangements are being structured.
- BOUNDARY
- The initiative is not a commercially available product. Its planned Q4 2026 start remains subject to consortium formation, team readiness, funding and programme definition.
03RESEARCH ARCHITECTUREREPEATABLE SYSTEM
03 / Research architecture
From domain evidence to controlled delivery.
ATP’s research programmes use a repeatable architecture. The components change by domain, but the discipline remains consistent.
07
Applications and controlled delivery
Connect validated intelligence to workflows, applications, APIs and infrastructure appropriate to the deployment environment.
06
Specialist evaluation
Evaluate answers, sources, behaviour and domain relevance with qualified specialists before broader use.
05
Models and specialised agents
Develop models and agents around defined domain tasks, evidence requirements and human decision contexts.
04
Retrieval and deterministic services
Connect evidence retrieval with calculations, rules, methods and services that should not depend on generative inference alone.
03
Governed corpus
Curate scientific, technical, regulatory and operational sources with provenance, scope and access boundaries.
02
Knowledge architecture
Structure the taxonomies, concepts, relationships, standards and source hierarchy that govern the domain.
01
Domain framing
Define the physical problem, operating context, decisions, methods, users and institutional responsibilities.
DOMAIN EVIDENCE → KNOWLEDGE ARCHITECTURE → GOVERNED INTELLIGENCE → SPECIALIST EVALUATION → CONTROLLED DELIVERY
A working architecture or prototype is not represented as a validated operational result. Public performance claims follow the relevant evaluation, deployment and measurement stages.
04TECHNICAL EVIDENCEINSPECTABLE ASSETS
04 / Technical evidence
The research system is made of inspectable assets.
ATP’s defensibility does not come from a list of future products. It comes from the knowledge structures, retrieval environments, evaluation systems, programme architectures and controlled infrastructure that can be reused across domain-intelligence programmes.
The registry below distinguishes foundations already built, working environments, assets in specialist validation, designed architectures and strategic infrastructure. These states do not imply equivalent maturity or commercial availability.
ASSETCLASSSTATEEVIDENCE
Forestry knowledge architecture
Domain knowledge systemBuilt foundationA hierarchical structure organising forestry domains, themes, subthemes and source relationships for TOMI AI Forestry.The architecture provides the governed foundation for development. It does not by itself represent a completed or validated model.
Forestry retrieval environment
Retrieval and evidence infrastructureWorking environmentA functional retrieval environment connecting forestry questions to indexed sources and supporting evidence.A working retrieval environment is not represented as complete domain coverage, specialist acceptance or production availability.
Answer-and-source evaluation set
Evaluation assetIn specialist validationAn initial set of forestry questions, responses and associated sources prepared for specialist review.The evaluation set remains part of the development process. No public accuracy or acceptance benchmark is claimed at this stage.
Domain-agent architecture
Model and agent systemIn developmentA reusable architectural direction for connecting specialised agents to governed retrieval, deterministic services and domain workflows.Agent scope, autonomy, performance and operational availability remain programme-specific and subject to evaluation.
GRID programme architecture
Climate-to-finance programme frameworkDesigned frameworkA structured pathway connecting industrial diagnosis, climate risk, applied R&D, transition planning, project preparation, sustainable finance and measurement.The framework is under programme structuring. It does not represent confirmed funding, participants, projects or measured results.
OLEA technical architecture
Domain-transfer architectureProposed architectureA concept and technical proposal applying ATP’s vertical-AI method to olive-sector knowledge, specialised agents and future field-data integration.The architecture is proposed rather than implemented. Funding, consortium, datasets, pilots and technical results remain unconfirmed.
KENE infrastructure layer
Controlled AI infrastructureStrategic infrastructureA common infrastructure direction for moving specialised models from governed development environments toward controlled delivery.Capacity, performance, geographic footprint and standard deployment configurations are not represented as universally available.
The ATP Domain Intelligence Stack
Seven connected layers organise delivery, intelligence, governed knowledge and controlled infrastructure.
07Applications and APIsWORKFLOWS · INTERFACES · CONTROLLED DELIVERY
06Specialised agentsTASK ORCHESTRATION · EVIDENCE USE
05Domain modelsDOMAIN REASONING · SPECIALIST CONTEXT
04Retrieval + deterministic servicesRULES · METHODS · CALCULATIONS
03Knowledge architectureTAXONOMIES · RELATIONSHIPS · STANDARDS
02Governed corpora and operational dataPROVENANCE · SCOPE · ACCESS
01KENE — controlled AI infrastructureDEVELOPMENT · EVALUATION · SERVING
CONTROL PLANE
01 Specialist evaluation
02 Provenance control
03 Deployment boundary
↕ATP develops domain intelligence as a connected system. Applications are visible at the top of the stack, but defensibility is created across the knowledge, evidence, evaluation and infrastructure layers below them.
Evaluation and Validation Loop
Validation is a system, not a final checkbox. Each programme defines evaluation criteria according to its domain, decision context and maturity stage.
Source ingestion → Provenance control → Retrieval tests → Domain evaluation → Specialist review → Failure analysis → Model or agent iteration → Controlled release
01Source ingestionINPUT CONTROL
02Provenance controlSOURCE GATE
03Retrieval testsEVIDENCE GATE
04Domain evaluationMETHOD GATE
05Specialist reviewHUMAN GATE
06Failure analysisERROR TAXONOMY
07Model / agent iterationSYSTEM REVISION
08Controlled releaseRELEASE GATE
CENTRAL CONTROLSPECIALIST VALIDATIONEvidence · Method · Behaviour
Domain accuracySource traceabilityRetrieval qualityMethod complianceCitation completenessAbstention behaviourSpecialist acceptanceDeployment readiness
Programmes on a common infrastructure direction
Separate domain programmes connect to one controlled infrastructure direction without becoming a fourth corporate pillar.
P.01TOMI AI ForestryIN DEVELOPMENTP.02TOMI AI Native ForestsUNDER STRUCTURINGP.03OLEA AIUNDER STRUCTURINGP.04GRIDUNDER STRUCTURINGP.05Future domain systemsFUTURE INTERFACE
PROGRAMME BUS→
CONTROLLED INFRASTRUCTUREKENETHE SOVEREIGN AI FACTORY
K.01Model environmentK.02Data boundaryK.03Delivery path
KENE is the cross-programme infrastructure direction through which ATP intends to support specialised model development and controlled delivery. It is not a fourth corporate pillar.
The presence of an asset in this registry indicates a defined and evidence-backed development state. It does not imply that every component is complete, validated, commercially available or deployed at scale.
05STRATEGIC INFRASTRUCTUREKENE
05 / Strategic infrastructure
Controlled infrastructure is part of the research method.
Domain intelligence is shaped not only by models, but also by where knowledge is processed, how evidence is governed and how systems are deployed. KENE — The Sovereign AI Factory is ATP’s strategic infrastructure layer for that transition.
KENE is intended to support the development, evaluation and controlled delivery of specialised AI across ATP programmes. It provides a common infrastructure direction for environments where privacy, institutional control, regional deployment and model governance matter.
A controlled path for developing, evaluating and serving specialised models and agents.
Deployment architecture designed around the data, access and regional requirements defined for each engagement.
A bridge from governed research environments to private, institutional or programme-specific applications.
KENE is strategic infrastructure, not a fourth corporate pillar. Its inclusion here does not represent universal availability, capacity, performance or a standard deployment configuration.
06MATURITY DISCIPLINESTATUS IS ARCHITECTURE
06 / Maturity discipline
Maturity is part of the architecture.
ATP uses explicit status language to distinguish what can be used today from what is being developed, structured or explored.
A system with a current delivery or commercial route. Product-specific scope and conditions remain on the relevant product page.
Active technical work supported by current project evidence, but not represented as generally available.
A programme with a defined problem and architecture whose funding, governance, participants or execution model are still being established.
A collaborative initiative that depends on formal consortium, team, funding and programme arrangements before execution.
A cross-programme technology or infrastructure direction supporting development and controlled delivery, without implying universal commercial availability.
A defined research direction without sufficient current execution evidence for a detailed public programme description.
RESEARCH → EVALUATION → CONTROLLED DEPLOYMENT → OPERATIONAL ACTION → MEASURED OUTCOME
ATP can document the intelligence, architecture and delivery environment it builds. Technical feasibility, implementation, verification, certification, financing and measured outcomes remain project-specific responsibilities.
07INSTITUTIONAL ROLESRESPONSIBILITIES EXPLICIT
07 / Institutional roles
Research advances when responsibilities remain explicit.
ATP develops domain intelligence through programmes that may connect scientific institutions, domain specialists, industrial organisations, infrastructure providers and funding mechanisms. Participation does not automatically imply endorsement, validation or commercial availability.
Amazonia Tech Partners
Domain-intelligence architecture, AI engineering, programme technology, applications and controlled delivery environments.
Universities and research institutions
Scientific knowledge, research methods, data curation, specialist evaluation and programme-specific technical work.
Industrial and domain organisations
Operational problems, use cases, domain constraints, implementation environments and project-specific validation.
Programme and funding institutions
Eligible funding instruments, programme governance and support according to formal agreements and applicable rules.
Independent professionals and verification bodies
Technical feasibility, verification, certification, assurance and other responsibilities that require independent professional judgement.
Infrastructure providers
Colocation, cloud capacity, hardware and other infrastructure services used within defined delivery architectures.
08BUILD WITH ATPCOLLABORATION PATHWAYS
08 / Build with ATP
Develop a domain-intelligence programme with ATP.
ATP works with organisations that hold domain knowledge, operational problems, scientific evidence or controlled-infrastructure requirements. Engagement begins by defining the problem, evidence, institutional roles and maturity boundary.
Forestry and nature intelligence
Develop governed intelligence for forestry, native forests, biodiversity and environmental decision contexts.
Green industry and climate finance
Structure programmes connecting emissions, climate risk, industrial transition, project preparation and sustainable finance.
Sovereign and controlled AI
Design specialised models and deployment environments for institutions with data, regional or infrastructure constraints.
Research to delivery
Build the next domain-intelligence system from the evidence outward.
Bring the domain problem, scientific knowledge, operational context or infrastructure requirement. ATP will help define the architecture, maturity path and controlled route to delivery.