Asset And Entity Relationship Querying With Natural Language Filters

via “advanced search filtering with temporal and entity extraction”

Hi HN,I built an open-source AI agent that has already indexed and can search the entire Epstein files, roughly 100M words of publicly released documents.The goal was simple: make a large, messy corpus of PDFs and text files immediately searchable in a precise way, without relying on keyword search

Unique: Combines NER with temporal filtering specifically for investigative workflows, likely building a knowledge graph of entity relationships extracted from documents rather than relying on external databases

vs others: More powerful than simple keyword filtering because it understands entity relationships and temporal context, enabling complex queries like 'all meetings between X and Y in Q3 2015'

via “database query and odata request generation”

Model Context Protocol (MCP) server for AI-assisted development of CAP applications.

Unique: Generates queries that respect CAP's entity model and CQL syntax — understands associations, compositions, and CAP-specific query semantics rather than generic SQL generation.

vs others: Produces CAP-native queries (CQL/OData) that integrate seamlessly with CAP's data layer, unlike generic SQL generators that would require translation or custom adapters.

** - The ThingsBoard MCP Server provides a natural language interface for LLMs and AI agents to interact with your ThingsBoard IoT platform.

Unique: Implements a dedicated Entity Data Query (EDQ) and Entity Count Query (ECQ) system with support for multiple filter types (equality, range, text search, regex) and a query builder pattern that constructs REST API payloads dynamically based on natural language intent, with built-in pagination and sorting support

vs others: Provides natural language entity querying (vs SQL or REST API syntax) with sophisticated filtering capabilities and relationship traversal, enabling non-technical users to perform complex data analysis without database knowledge

via “field-value-filtering-and-search”

** - Perform queries and entity operations in your [Fibery](https://fibery.io) workspace.

Unique: Exposes Fibery filtering as MCP tool, allowing agents to construct queries with field-level filters without writing GraphQL. Supports multiple filter operators (equals, range, text search) and boolean combinations, enabling flexible entity queries.

vs others: Agents can filter entities efficiently without fetching full collections; direct API clients require agents to construct where clauses manually or fetch all entities and filter in-memory, reducing efficiency.

via “natural language project search and filtering”

Unique: Adds conversational search to project management interface rather than requiring users to learn structured filter syntax, but likely uses simpler pattern matching than semantic search tools, limiting query complexity and ambiguity handling

vs others: More intuitive than structured filters in Monday.com or Asana, but less powerful than semantic search in Notion or Slack which use embeddings for fuzzy matching

via “asset search and discovery with semantic filtering”

Unique: Combines full-text search with semantic similarity matching, allowing users to find assets using natural language descriptions that don't exactly match indexed keywords (e.g., 'portable computer' matches 'laptop')

vs others: Provides semantic search for asset discovery, whereas traditional asset management systems rely on exact keyword matching and require users to know precise asset naming conventions

via “entity-relationship-inference-from-text”

Unique: Performs bidirectional entity-relationship inference — extracting both explicit relationships mentioned in text and inferring implicit associations through linguistic patterns (e.g., possessive constructions, verb phrases indicating ownership or composition)

vs others: More automated than manual ER diagramming tools but less precise than structured schema specification languages because it relies on natural language ambiguity resolution rather than explicit syntax

via “asset-search-and-filtering”