Collibra

Collibra Unstructured Data for AI is a Collibra solution for preparing and governing unstructured document data for AI use cases. It supports document curation, AI-assisted taxonomy and metadata tagging with evidence and confidence scores, sensitive-data detection, and quality checks for duplicates, conflicts, and stale content. Data slices can be refreshed on a schedule so AI-ready datasets remain current. Collibra describes the solution as providing governance for AI use cases, models, data, and agents with end-to-end traceability, oversight, and compliance.

Deasy Labs is an enterprise platform for curating unstructured organizational data for AI applications. It connects to sources such as SharePoint and S3, OCRs, parses, and chunks files, then builds taxonomies, applies metadata, detects sensitive information, and evaluates file quality and relevance for a stated use case. Users can create data slices by topic, time, quality, relevance, or sensitivity and deliver them to RAG pipelines, search systems, and AI agents, while writing metadata back to source systems. Deasy continuously monitors connected sources, enriches new content, and refreshes datasets to reduce the effects of stale, duplicated, conflicting, irrelevant, or sensitive files. The platform provides APIs and a Python SDK, supports deployment in a customer's own environment, and allows use of customer-provided models and LLM endpoints. The company was acquired by Collibra, according to the video description.

Deasy Labs is an AI data-curation platform with APIs and a Python SDK for preparing unstructured organizational data for RAG systems, search, and AI agents. It connects to sources such as SharePoint and S3, OCRs, parses, chunks, and normalizes files, then builds taxonomies, applies metadata, detects sensitive data, and evaluates file quality and relevance against a use case. Users can slice the enriched corpus by relevance, topic, time, quality, or sensitivity, write metadata back to source systems, or deliver datasets to downstream retrieval pipelines. Data slices are continuously monitored and refreshed as source content changes. The SDK can also generate context files and folder indexes for coding agents, as demonstrated in the video.

Colibrì is a pure-C inference engine and open research platform for running large mixture-of-experts models on consumer and heterogeneous hardware without engine dependencies. It treats VRAM, system RAM, and storage as a unified multitier inference hierarchy, streaming expert weights from disk and coordinating placement, storage I/O, scheduling, kernels, speculation, and CPU/GPU overlap. The project provides chat, server, and web front ends, with support for CPU, CUDA, Metal, and dual-SSD streaming; its design rules prohibit silently changing model precision or router semantics when fast memory is insufficient.

Fluree is an enterprise AI data platform that turns raw, structured, and unstructured data into trusted, queryable knowledge graphs. Its platform uses governed data models, taxonomies, golden records, entity resolution, semantic layers, and GraphRAG-powered retrieval to prepare enterprise data for conversational analytics, enterprise search, decision intelligence, and AI agents. Users can ask questions, create dashboards, build applications, and deploy governed agents over the resulting knowledge graph.