Hosted search
for humans
and machines.

Autocompletion for your users. MCP for their agents. One hosted index serves both.

GET /v1/search

$ curl 'https://api.eztype.io/v1/search?service=eztype-demo&index=countries&q=sw'

200 OK8 ms

{
  "suggestions": [
    { "title": "Sweden",       "weight": 10402070 },
    { "title": "Switzerland",  "weight":  8654622 },
    { "title": "Swaziland",    "weight":  1136281 }
  ]
}

For humans

Autocompletion.

Sub-50ms prefix matching across four demo indexes. Pick one and start typing.

195 countries, geo-rankedGET /v1/search?index=countries

For machines

MCP.

Same indexes — now queried by Claude through an MCP endpoint. See the reasoning, the tool calls, and the results in one trace.

Pick a prompt below — or type your own — to start.

Anonymous and powered by Claude + the EZType MCP service.

11regions

Worldwide, latency-selected

<50ms

Typical end-to-end query

3protocols

Autocomplete, full-text, MCP — same index

What's in the box

Everything modern search needs.

MCP-native

Point Claude Desktop or any MCP client at your index. Read-only, tenant-scoped, with the same versioning guarantees as the REST API.

Versioned indexes

Every change is a versioned artifact. Roll back to any prior version in one call. Pin AI workflows to a specific version for determinism and audit.

Multi-protocol search

One index, three ways to query. Sub-100ms typeahead for autocomplete, full-text REST for richer queries, and MCP for AI tools — backed by the same data.

Delta-encoded updates

Replace the index, update a single record, delete a record — all the same model. Per-record updates ride a delta log; coalesce folds them into a fresh base on a schedule.

Global multi-region

Place indexes in any of Fremont, Newark, Atlanta, Dallas, London, Frankfurt, Toronto, Sydney, Tokyo, or Mumbai.

Latency-based routing

The region closest to your end user, by latency, is automatically selected for each query — with cross-region redundancy on failure.

High availability

Redundant search engines within each region plus automatic cross-region failover. S3 is the durable log of record.

Per-tenant clusters

Every paying tenant gets a dedicated OpenSearch cluster in its own namespace. No noisy-neighbor surprises, no shared indexes.

Linear scalability

Scale search indexes linearly to meet virtually any requirement.

Geosearch

Include GPS coordinates in your dataset to rank results by distance from the user.

Weighted search

Add integer weights to your dataset to boost the ranking of more important records.

Fast uploads

Push your dataset at full network speed. Multi-GB files supported.

Hosted searchfor humansand machines.

Autocompletion.

MCP.

Everything modern search needs.

Hosted search
for humans
and machines.