MOLI is a platform for Molecular Intelligence.
Molecular Intelligence emerges from combining scientific context, molecular modeling, and scientific reasoning to understand, investigate, learn about, and create molecular systems.
MOLI is not an LLM and does not require an AI agent to perform reproducible science. Human scientists can operate the same scientific infrastructure directly, while MOLI Agent may augment reasoning and agency where useful.
The platform is designed to be local-first and remote-ready: scientific semantics remain independent of deployment topology so the same architecture can evolve from local workflows to shared, distributed, or hybrid infrastructure.
MOLI
platform / umbrella
│
┌──────────────────┼──────────────────┐
│ │ │
▼ ▼ ▼
Scientific Context MolSysSuite MOLI Agent
│ │
┌────┼────┐ ├── MolSysMT
▼ ▼ ▼ ├── MolSysViewer
Sabueso Praxis Nextia ├── TopoMT
├── ...
└── MolSys-AI
The diagram shows composition, not mandatory execution flow.
Scientific Context is the conceptual grouping of three complementary forms of persistent scientific context:
- Sabueso — Knowledge: what is known about molecular entities, systems, properties, relationships, and observations, preserving traceable source assertions.
- Praxis — Know-how: what scientific tasks we know how to perform reproducibly, through reusable Capabilities and Protocols.
- Nextia — Discovery: what we are trying to discover, what we have tried, observed, learned, rejected, and decided.
Scientific Context is an architectural concept, not a requirement for a separate package or governance repository.
MolSysSuite is MOLI's molecular modeling ecosystem. Its components provide molecular-system representation, interoperability, computation, simulation, analysis, and visualization.
MolSysSuite is a first-class MOLI component with its own delegated internal governance. MOLI governs MolSysSuite at the platform boundary; MolSysSuite governs its internal components and shared modeling-ecosystem policies.
MolSys-AI belongs to the MolSysSuite domain as the AI subsystem specialized in understanding and operating MolSysSuite.
MOLI Agent is the optional scientific reasoning and agency component operating across Scientific Context and MolSysSuite.
It may help interpret scientific intent, assemble relevant context, reason over DiscoveryProjects, use methodological Capabilities, delegate modeling-specialist work to MolSys-AI Agent, invoke authorized scientific tools, and interpret results while preserving explicit authority and approval boundaries.
MOLI Agent is not the MOLI platform itself.
Composition does not imply isolation. Scientific Context and MolSysSuite may interoperate directly where appropriate, while MOLI Agent may reason and act across both.
MOLI Agent
│
reasoning / agency
│
┌───────────┴───────────┐
▼ ▼
Scientific Context ◄──────► MolSysSuite
│
┌────┼────┐
▼ ▼ ▼
Sabueso Praxis Nextia
Ownership of a scientific concept does not imply isolation, and conceptual interoperability does not require circular package dependencies.
MOLI can be understood through three compatible architectural views:
- Structural: Scientific Context ↔ MolSysSuite, with optional MOLI Agent across both.
- Functional: Knowledge | Modeling | Capabilities | Discovery.
- Dynamic: KNOW → MODEL → DO → DISCOVER → LEARN → KNOW.
Knowledge Modeling Capabilities Discovery
│ │ │ │
Sabueso MolSysSuite Praxis Nextia
The dynamic view makes learning explicit: discovery experience may improve Knowledge and Know-how through controlled curation and validation. Promotion does not imply publication.
Molecular Intelligence is a property of the integrated scientific system, not the name of a subsystem and not a synonym for AI:
Molecular Intelligence = Scientific Context + Molecular Modeling + Scientific Reasoning
Reasoning may be human, agent-assisted, or a combination of both.
MOLI may eventually compose the scientific state held across Sabueso, Praxis, Nextia, and MolSysSuite into traceable ProjectBriefings, ProgressBriefs, and ProjectReports. These structured communication artifacts may then be rendered as written dossiers, interactive reports, slides, or narrated video while preserving provenance, uncertainty, confidentiality, and the ownership of the underlying scientific objects.
See devguide/SCIENTIFIC_COMMUNICATION.md.
The frozen conceptual baseline is documented in architecture_1.0/.
It defines component boundaries, Scientific Context Assembly, the learning loop, epistemic distinctions, agent specialization, object identity and portability, visibility/confidentiality, deployment independence, examples, diagrams, and scientific/operational stress tests.
Architecture and implementation evolve at different rates: implementation repositories may mature without silently changing the frozen conceptual meanings.
MOLI also acts as the governance and coordination repository for platform-level contracts.
moli.tomlis the machine-readable registry of MOLI components and governance relationships.MOLI_GUIDE.mdis the concise component-facing governance guide.devguide/contains durable governance and development knowledge.devguide/INFRASTRUCTURE_CENSUS.mdrecords observed implementation maturity and pending platform infrastructure.devguide/governance/policy_inheritance.mddefines how delegated ecosystems inherit MOLI engineering policy.
The governing rule is:
A concern is governed at the lowest level that owns the shared contract it affects.
Sabueso, Praxis, Nextia, and MOLI Agent are directly coordinated through MOLI for shared platform contracts. MolSysSuite is also a MOLI component, while governance of its internal ecosystem is delegated to the MolSysSuite repository.
MOLI
├── uibcdf/moli
├── uibcdf/sabueso
├── uibcdf/praxis
├── uibcdf/nextia
├── uibcdf/moli-agent
└── uibcdf/molsyssuite
└── modeling components and MolSys-AI
Real discovery programs may remain private while using the public MOLI infrastructure. Open-source software does not imply publication of DiscoveryProjects, proprietary Know-how, Evidence, Candidates, or molecular assets.