Grounded answers
Ask questions over THiNK-managed knowledge bases and get answers supported by retrieved evidence, graph context, and provenance.
SemanticMatter Sculpin Agents
SculpinKnowledge-aware AI agents for THiNK
Sculpin connects AI agents to curated knowledge bases, semantic metadata, ontologies, tools, and graph evidence—so users can ask questions, explore relationships, and act on trusted information without losing provenance. It is built for teams that need AI grounded in their own documentation and data models, not generic answers detached from context.
Graph-aware reasoning
Let agents follow semantic relationships across datasets, documents, concepts, procedures, and resources.
Action-ready intelligence
Connect agents to tools, workflows, and APIs so knowledge can move from explanation to execution.
From semantic documentation to intelligent assistance
THiNK provides the semantic foundation: structured metadata, ontology alignment, provenance, access control, and reusable knowledge models. Sculpin builds on that foundation by giving users a natural way to interact with it.
Instead of searching across disconnected files, dashboards, and systems, users can ask an agent. The agent retrieves relevant knowledge, inspects relationships in the graph, cites its evidence, and guides the user toward the next step. Sculpin does not replace the knowledge graph—it makes the knowledge graph accessible.
Core message
THiNK gives organizations a trusted semantic foundation Sculpin makes that foundation accessible, actionable, and intelligent through knowledge-aware AI agents.
What Sculpin adds to THiNK
Agents that stay scoped to your governed knowledge, ground their answers in evidence, and connect to the tools your teams already use.
Knowledge-base aware assistants
Create agents that answer from selected THiNK knowledge bases and documentation collections, scoped to the right domain, team, or operational context.
Evidence-first responses
Agents ground answers in retrieved content and graph evidence, so users see not only the answer, but where it came from.
Graph-guided discovery
Move beyond keyword search by following links between concepts, datasets, procedures, software, people, projects, and provenance records.
Tool-connected workflows
Connect agents to external tools and services—domain APIs, document processors, diagram and chemistry tools, and graph query engines.
Human-reviewed automation
Agents propose summaries, metadata, mappings, or next actions while experts stay in control of validation and publication.
Open integration model
Expose agents through an OpenAI-compatible API pattern to fit existing applications, portals, enterprise chat, and developer tooling.
How Sculpin works
A grounded, provenance-aware, tool-capable workflow—from a question to a trusted, actionable response.
Ask
A user asks a question in natural language, scoped to a selected agent and knowledge base.
Retrieve
The agent retrieves relevant documents, metadata, ontology terms, graph relationships, and provenance.
Reason
Retrieved evidence is combined with the semantic structure of THiNK to produce a focused response.
Cite
The answer includes evidence, references, or provenance so users can inspect the basis for it.
Act
When relevant, the agent calls tools, generates artifacts, queries the graph, or supports the next step.
Improve
Validated outputs and user feedback strengthen the underlying knowledge base over time.
Priority use cases
Enterprise knowledge assistants
Give teams a trusted assistant for internal documentation, procedures, reports, and domain knowledge.
Semantic search and discovery
Help users find relevant datasets, documents, concepts, and relationships without knowing exact terminology.
Data documentation support
Assist data stewards and researchers with metadata extraction, provenance capture, and FAIR-aligned documentation.
Procedure and compliance support
Answer questions from controlled procedures and governing documents while preserving traceability to the source.
Knowledge graph investigation
Explore how datasets, methods, software, instruments, projects, and people connect across the graph.
AI-enabled technical workflows
Connect THiNK knowledge to external services so users can generate diagrams, summaries, mappings, and structured outputs.
THiNK and Sculpin together
THiNK
The governed semantic layer
The catalog, metadata model, ontologies, mappings, provenance, and access structure that keep knowledge trustworthy and reusable.
Sculpin
The agentic interaction layer
The agents, tools, retrieval strategies, and conversational workflows that make that knowledge usable in daily work.
What changes in practice
Static knowledge bases → Interactive assistants
Document search → Evidence-grounded answers
Ontologies → Natural language exploration
Metadata → Actionable guidance
Provenance → Trustworthy AI responses
Expert-only graph access → Usable knowledge for every team
Together, THiNK and Sculpin let organizations document knowledge once, govern it properly, and reuse it through trusted AI assistance—turning a semantic knowledge base into grounded, domain-specific assistants for your teams.