PATH 01
Instant Search
Pre-indexed, sub-100ms retrieval for high-frequency queries over stable corpora. High-frequency query patterns are pre-indexed using a KV-cache architecture. Answers appear before the user finishes typing.
Chordian Search
Search that reasons. Not search that retrieves.
Chordian Search is an AI-agent-based enterprise search engine. Unlike conventional search, which returns a ranked list of documents, Chordian Search dispatches an autonomous retrieval agent that runs seven parallel retrieval paths, fuses the results using Reciprocal Rank Fusion (RRF), and continues reasoning — re-querying, cross-referencing, and refining — until it reaches a confident, source-attributed answer. Every result is structured JSON with full provenance. Every search query also writes back to your private Chordian Memory, making your institutional knowledge richer with each use.
Retrieval paths
Seven paths. One fused answer.
PATH 02
Semantic Search
Meaning-based retrieval using embedding models. Finds conceptually related content even when the exact words don't appear in the query. Configurable per deployment.
PATH 03
Keyword Search
BM25 exact-match precision for proper nouns, product names, contract numbers, and regulatory references. Always runs in parallel with semantic search.
PATH 04
Graph Search
Relationship traversal across your private institutional memory. Who owns what, who knows whom — the answer is in the edge, not the node.
PATH 05
GraphRAG
Graph-augmented generation grounded in your private memory. Higher accuracy than pure vector retrieval on relational questions.
PATH 06
Web Intelligence
Real-time open web enrichment via the Chordian Agentic Crawler. Live external signals integrated with your private memory. Always attributed to source.
PATH 07
Deep Search (Agentic Retrieval)
An autonomous AI retrieval agent that re-queries, cross-references, and reasons across multiple hops until it reaches a confident answer. Returns structured JSON with every reasoning step attributed.
The fusion layer
All seven paths run in parallel. Reciprocal Rank Fusion combines their ranked outputs into one final result — with source attribution, confidence scores, and retrieval path transparency. No black box. Every answer is fully traceable.
Why fewer tokens
Chordian Search delivers pre-structured, source-attributed, confidence-scored context directly to your LLM. Your model spends its token budget on reasoning — not on re-learning what your organisation already knows.
API
Single-line query. Structured JSON response. Source attribution on every result. See full API reference in the developer documentation.