Keyword Search
Matches precise words and phrases.
CAPABILITIES
Technologies and methods
Knowtive chooses technology based on the problem — not the other way around.
A simple solution may be the right solution. Other information problems require a combination of search, knowledge structure, AI, integrations and governance.
This is a capability map — not a fixed vendor stack.
01Sources→
02Retrieval→
03Knowledge structure→
04AI→
05Governance→
06Use
01 · RETRIEVAL
Retrieval and search
Retrieval determines which information an AI system sees. Good retrieval can therefore matter more than simply choosing a larger language model.
Read about retrieval in a Knowledge Assistant →Semantic Search
Finds content by meaning and context.
Vector Search
Searches numerical representations of content.
Hybrid Search
Combines keyword and vector search.
Embeddings
Represent semantic characteristics numerically.
Reranking
Reprioritises the strongest retrieved results.
02 · KNOWLEDGE STRUCTURE
Knowledge structure
Metadata describes information. Taxonomies organise concepts. Ontologies describe meaning and relationships. Knowledge graphs represent specific entities and connections.
Read about Knowledge Architecture for AI →Metadata
Provides context and filtering.
Taxonomies
Create a controlled vocabulary.
Ontologies
Describe relationships and semantic knowledge.
Entity Models
Define central entity types.
Entity Resolution
Connects name variants to one identity.
Knowledge Graphs
Represent entities and semantic relationships.
Information Architecture
Organises content around user needs.
03 · AI & MODELS
AI and models
Model choice depends on the problem, security, data privacy, performance, cost and hosting requirements. The choice of AI model is part of the architecture — not a prerequisite.
See how models fit the assistant architecture →Retrieval-Augmented Generation (RAG)
Grounds generated answers in retrieved information.
Large Language Models (LLM)
Interpret and generate language in context.
AI Assistants
Support information tasks through a usable interface.
Enterprise Chatbots
Combine conversation with governed organisational knowledge.
AI Agents
Coordinate bounded tools and process steps.
Cloud Models
Operate in cloud environments with selected controls.
Private AI
Can be adapted to data-handling requirements.
Local Models
May run locally when requirements and trade-offs support it.
04 · INTEGRATION
Integration and data sources
Useful knowledge systems often need to connect existing sources rather than create another isolated silo. The architecture depends on the systems and interfaces actually available.
See external sources in Research & Surveillance →APIs
Connect systems through controlled interfaces.
SharePoint
A general example of an internal document source.
Document Repositories
Contain files, metadata and versions.
Databases
Provide structured data.
External Information Sources
Provide relevant records and signals.
Document Ingestion
Retrieves, normalises and prepares documents.
05 · GOVERNANCE & TRUST
Governance, security and trust
An enterprise AI system should not only produce an answer. It should make it possible to understand what information was used, whether access was permitted, where the answer originated and how quality can be evaluated.
Read about monitoring with provenance →Access Control
Restricts access to information.
Permissions
Apply rights within the selected architecture.
Data Privacy
Informs processing, model and hosting choices.
Data Governance
Defines responsibility and rules for data use.
Data Residency
May constrain where data is processed and stored.
Source Provenance
Preserves documented origin.
Source Citations
Link answers back to evidence.
Evaluation
Measures retrieval and answer quality.
Auditability
Supports review where infrastructure allows it.
PRAGMATIC ARCHITECTURE
Not every problem needs the same architecture
01 · SIMPLE
Small document collection
Clean metadata + good search
02 · ENTERPRISE
Knowledge Assistant
Hybrid retrieval + RAG + access control + evaluation
03 · COMPLEX DOMAIN
Research knowledge
Ontologies + entity resolution + knowledge graphs
Knowtive makes knowledge useful.
Talk to Knowtive about the right architecture
Start with the problem. Together, we can identify the combination of structure, search and AI that makes sense.
Let’s talk