Orbifold Consulting

@orbifold
1 Followers
2 Following
74 Posts
Graph AI consulting and intelligence.

SurrealDB v3 solves the graph RAG ugliness of a vector DB here, a graph DB there, a document store somewhere else. With native vectors, documents, and graph relations all in one engine, SurrealDB v3 finally gives a unified substrate for #GraphRAG. One query language. One data model. One place to store chunks, embeddings, entities, and edges.
But as an all‑in‑one platform for building #KnowledgeGraphs, it’s one of the most compelling options available today.

https://surrealdb.com/3.0

Zvec by Alibaba is a lightweight, lightning-fast, in-process vector database. Beats any other vector db out there. https://zvec.org/en/ #VectorDatabase
OpenAI acquired the claw but there are various alternatives and even some with semantic graph brains. MemU has semantic linking, multi-hop reasoning, a unified multi-modal memory. Not using a graph database though. Am sure vendors are figuring out how to catch this train, won't take long.
#GraphAI
https://memu.pro/memory-graph-agent
Memory for Proactive 24/7 Agents | MemU - Let AI Agents Work for You

Always-on memory for proactive AI agents. Predict user intentions, act proactively, work for you 24/7 — even while you sleep.

MemU
The Knwler hairball graph visualization (Cytoscape) has been replaced by a custom layout algorithm and SVG rendering.The layout now depicts much more clearly the relationships. The concentric circles and fading edges (proportional to distance) are organized according to degree and minimize crossings.
Hope you like it.
https://github.com/Orbifold/knwler
#GraphViz #KnowledgeGraphs

Memgraph 3.8 is here with real‑time graph intelligence.

Three things stand out:
- Atomic GraphRAG: graph augmented questions with one query.
- Vector Search in the Same Store: new hybrid graph‑vector patterns
- Parallel Runtime Improvements: analytical and traversal-heavy queries now run faster thanks to deeper parallelization across the runtime.

https://memgraph.com/blog/memgraph-3-8-release-atomic-graphrag-vector-single-store-parallel-runtime
#GraphDatabase #GraphRAG #KnowledgeGraphs

Anthropic knowledge-work plugin for in-house legal teams, primarily designed for Cowork. It automates contract review, NDA triage, compliance workflows, legal briefings, and templated responses -- all configurable to your organization's specific playbook and risk tolerances.
#LegalAI #LegalTech
https://github.com/anthropics/knowledge-work-plugins?tab=readme-ov-file
GraIL is a graph neural network for inductive knowledge graph reasoning. Unlike embedding-based approaches, GraIL model is able to predict relations between nodes that were unseen during training and achieves state-of-the-art results in this inductive setting. https://github.com/kkteru/grail #KnowledgeGraphs #GraphMachineLearning
KGTK is a Python library for easy manipulation with #KnowledgeGraphs. It provides a flexible framework that allows chaining of common graph operations, such as: extraction of subgraphs, filtering, computation of graph metrics, validation, cleaning, generating embeddings, and more. https://kgtk.readthedocs.io/en/latest/
Makes my day🙂 Graph ML identifies promising additional drug indications for diseases such as acute lymphoblastic leukemia and Alzheimer’s disease.
https://www.nature.com/articles/s41551-025-01598-z
#KnowledgeGraphs #GraphML #DrugRepurposing
Reconciliation of triple stores and property graphs is an old topic and many have tried to tackle it (from every angle and tech). It's polarizing and the latest RDF-* and SPARQL-* announcements turn an already complex field into more syntax and complexity. When flexibility equals complexity it seldom leads to more adoption.
#GraphDatabases #RDF #KnowledgeGraphs
https://olafhartig.blog.liu.se/
https://www.ontotext.com/knowledgehub/fundamentals/what-is-rdf-star/