- Graphwise Platform Documentation
- Graphwise Documentation
- Graphwise Release Notes
- GraphRAG 1.0.0 Release Notes
GraphRAG 1.0.0 Release Notes
18/02/2026
GraphRAG 1.0 delivers the intelligence layer that powers the Graphwise platform. It transforms unreliable AI into trusted enterprise intelligence through complete transparency and auditability by integrating knowledge graphs, semantic search, and conversational AI into a single architecture. Users get complete visibility into how answers are generated, what sources are used, and why specific conclusions are reached. No more guessing about accuracy of AI responses.
GraphRAG provides a unified platform that meets corporate standards for accuracy, auditability, and compliance. Each response is grounded in the semantic integrity of corporate data.
This unique architecture enables organizations to safely deploy GenAI for mission-critical decision-making, ensuring every answer is grounded in the semantic integrity of their corporate data.
GraphRAG improves accuracy across various query types by combining three retrieval methods:
Graph Retrieval: Relation-based queries using SPARQL over knowledge graphs
Vector Search: Semantic similarity matching using embeddings
Full-Text Search: Traditional keyword matching to ensure no relevant details are missed
This approach finds relevant information that single-method systems miss, especially for complex questions requiring multiple data connections.
User questions are automatically enriched using enterprise taxonomies and ontologies. The system automatically expands brief or unclear queries with relevant context, ensuring the AI understands what you are actually asking for.
GraphRAG provides complete visibility into how answers are produced.
Provenance and citations
Responses include provenance metadata and citations so users can verify claims.
Traceability of retrieval steps
Power users can inspect the specific tools and steps used (e.g., SPARQL queries, vector results) by the agentic framework and by reviewing the path the answer followed to reach the final result.
SKOS concept highlighting
When responses reference knowledge graph concepts, semantic terms are highlighted and clickable, exposing SKOS properties such as:
URIskos:PreLabelskos:altLabel(synonyms)skos:definition
Integrated guardrails (input + output)
Native guardrails monitor both stages and reject inappropriate, off-topic, or non-compliant content.
Role-based access control
Managed via Keycloak.
Multi-turn conversations with context retention
Suggested follow-up questions based on current discussion facilitating a deeper and more structured exploration of the data
All user sessions and conversation histories are securely stored, allowing users to easily revisit and reference past interactions
Real-time streaming responses using SSE (Server-Sent Events) with formatted content
GraphRAG 1.0 uses a decoupled architecture compatible with
Major LLM providers (e.g., OpenAI, Azure AI, AWS Bedrock)
Multiple vector databases (e.g., OpenSearch/Elasticsearch)
Various Identity Providers
External Secrets support for AWS Secrets Manager, Azure Key Vault, Google Secret Manager, and HashiCorp Vault
Multi-tenancy capabilities to separate concerns and isolate different teams and projects
This reduces vendor lock-in and lets IT change models or infrastructure as requirements change and evolve.
A workflow engine orchestrates core logic through dedicated workflows, such as concept extraction, expansion, and query and answer enrichment and supports
Rapid experimentation
Component chaining and reuse
Custom enterprise flows without modifying the core codebase
Sync & Async Querying
The API supports
Synchronous mode (immediate response)
Asynchronous mode (polling or streaming) for long-running queries without freezing the UI
Client & Browser Persistence
Persistent client IDs
GraphRAG 1.0 is preconfigured to support
AWS Bedrock: Titan (v1, v2), Cohere (English, Multilingual v3)
OpenAI: text-embedding-ada-002, text-embedding-3-small, text-embedding-3-large
PoolParty 9 or later
GraphDB 11.2 or later
GraphRAG Workflow Engine: n8n v2.4.4
PostgreSQL v17
Java Runtime: OpenJDK 21
Node.js 18 or later
Kubernetes (K8s) with cloud load balancers (e.g., AWS LB or Azure Application Gateway)