Remember

Ahura Search

High performance vector/embedding search

A full search engine, run for you. Index documents, records, logs, products or anything else, then query by keyword, phrase, filter, meaning or a blend of all of them. It is not limited to what your agents create: point it at any data and search it fast, with nothing to deploy, size or tune.

What it does

A real search engine
Relevance-ranked full-text search with phrases, filters, sorting and highlights.
Meaning as well as words
Store vectors with your documents and search by similarity, or blend keyword and vector results into one ranking.
Any data, any source
Index agent output, your own databases, files, logs and product catalogues. If it can be JSON, it can be searched.
Nothing to run
No cluster to deploy, size, secure or patch. Create an index and start querying.

An agent calls it like this

{ "tool": "search.query", "index": "policies", "q": "refund policy exceptions", "limit": 5 }

Interactive simulation

Try Ahura Search

Press Run to send the call. The dashboard shows what it would use in a partition. Simulated: nothing is actually executed.

ahura mcp · ptn_demo_7f3a

# Index a document on write.

→ POST /mcp

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "search.put",
    "arguments": {
      "index": "policies",
      "id": "p_17",
      "doc": {
        "title": "Refund policy",
        "body": "Refunds within 30 days, with exceptions for…"
      }
    }
  }
}

# Find it again.

→ POST /mcp

{
  "jsonrpc": "2.0",
  "id": 2,
  "method": "tools/call",
  "params": {
    "name": "search.query",
    "arguments": {
      "index": "policies",
      "q": "refund policy exceptions",
      "limit": 5
    }
  }
}

# Nearest by meaning.

→ POST /mcp

{
  "jsonrpc": "2.0",
  "id": 3,
  "method": "tools/call",
  "params": {
    "name": "search.knn",
    "arguments": {
      "index": "policies",
      "vector": "[…384 floats]",
      "k": 3
    }
  }
}

Ready. Press Run.

ptn_demo_7f3a

Partition · N. Virginia · created just now

Simulated

Activeus-eastRetention hold

days left

Run time

0.0s

0 calls

Size

2.3MB

0.00 ACU used

Services in this partition

Search
0 calls · 0.00 ACU

Spend by service

Ledger

No usage yet.

Why it matters

Where Ahura Search beats running a search cluster.

Search touches everything: documents, records, logs, products, memories. A search cluster is powerful but costly to run, size and tune.

A search cluster you run

Stand up, size and secure a cluster, with a baseline cost even when idle.
Build and maintain pipelines to feed it, and watch for it to fall behind.
Run keyword search and vector search as two separate systems.
Tune analysers, shards and relevance by hand as data grows.

Ahura Search

A managed, high-performance search engine. Create an index with one call.
Index on write, from agents or from any other source.
Keyword, vector and blended hybrid search over the same documents.
Sensible ranking, language handling and scaling out of the box.

Ranked
Relevance scoring with phrases, filters and highlights.
Semantic
Vector similarity with embeddings from Ahura AI.
Everything
Documents, records, logs and catalogues in one engine.
Cost
Billed for the storage an index uses, plus queries. Drop an index and the charge stops.

Search storage is not free. Indexes, including their vector data, are billed for the space they use as long as they exist, so expire or drop what you no longer need. Ahura Search finds things. It does not replace your system of record: keep your source of truth where it lives and search a copy here.

In depth

What Ahura Search can do.

A proven engine, managed for you
A high-performance search engine, run and scaled for you. You get its power without operating it.
Keyword search that ranks well
Results are ranked by relevance, rewarding rare terms and tighter matches.
Phrases, exclusions and fields
Search for exact phrases, exclude terms with a minus, or target a specific field.
Filters, ranges and sorting
Narrow by keyword, number or date fields and sort by score or by any field.
Highlights
Get matching snippets back with each hit, so a person or an agent can see why it matched.
Vector search
Store embeddings alongside documents and find the nearest by cosine, dot product or distance.
Hybrid search
Blend keyword and vector results into one ranking, with a dial between the two.
Type-ahead
Suggest completions from the terms in your own index.
Many indexes, one place
Keep separate indexes for separate data, each with its own fields and analysis, in the same partition.
Language handling
Unicode-aware text analysis with optional English stemming and stop words, chosen per field.
Scales with you
From a few documents to very large indexes, without re-platforming or capacity planning.
Storage is billed
An index is charged for the storage it uses, vectors included, plus a price per query. It costs money for as long as it exists, so expire or drop what you no longer need.

Use it for everything

Not just what your agents write.

Ahura Search is a general-purpose, high-performance search engine. Agents can use it, and so can the rest of your stack.

Agent retrieval and RAG
Give agents fast, relevant context from your documents, notes and past runs.
Site and app search
Power search boxes with ranked results, filters, highlights and type-ahead.
Product and catalogue search
Combine keyword relevance, filters and similarity for shopping and discovery.
Knowledge bases
Make policies, wikis, tickets and manuals searchable by words and by meaning.
Logs and events
Index operational events and find the needle with phrases, filters and time ranges.
Records and CRM data
Search customers, orders and cases that live in other systems.
Research and e-discovery
Find every mention across large document sets, and show where it appears.
Recommendations by similarity
Surface the nearest items to a query, a document or a user’s history.

Every call, listed

11 tools your agent can call. Each one is a standard MCP tool with a typed schema and a price shown in tools/list.

Indexes

search.index.create / get / list: Define an index and its fields: text, keyword, number, date or vector.
search.index.drop: Delete an index and everything in it.
search.stats: Document counts and index size.

Documents

search.put: Index a document, with an optional vector.
search.bulk: Index up to 50 documents at once.
search.get: Read a stored document.
search.delete: Remove a document and its index entries.

Querying

search.query: Keyword search with filters, sorting, paging and highlights.
search.knn: Nearest-neighbour search over vectors.
search.hybrid: Keyword and vector blended into one ranking.
search.suggest: Prefix completions from an index.

Search does not create embeddings itself. Use Ahura AI to embed text, and Ahura Search stores and compares the vectors.

Use cases with Search

All use cases

and 3 more

Industries using Ahura Search

Healthcare

Works with

Next: Ahura AI

Ahura Memory

Fast working memory for agents.

Ahura Store

Persistent files and objects.

Ahura NoSQL

Document collections without the setup.

Ahura Graph

Relationships an agent can walk.

Partitions

Where this service's data lives and how it freezes.

Ahura Search

What it does

An agent calls it like this

Try Ahura Search

Where Ahura Search beats running a search cluster.

A search cluster you run

Ahura Search

What Ahura Search can do.

A proven engine, managed for you

Keyword search that ranks well

Phrases, exclusions and fields

Filters, ranges and sorting

Highlights

Vector search

Hybrid search

Type-ahead

Many indexes, one place

Language handling

Scales with you

Storage is billed

Not just what your agents write.

Agent retrieval and RAG

Site and app search

Product and catalogue search

Knowledge bases

Logs and events

Records and CRM data

Research and e-discovery

Recommendations by similarity

Every call, listed

Indexes

Documents

Querying

Use cases with Search

Industries using Ahura Search

Works with