Search. Ask. Observe.

Pattern-recognition search that finds what keyword search misses, answers questions from your own content, and tells you what to tune.

Get started

Questions are answered by AI from this catalogue. How your text is handled.

Film data and posters from TMDB. This product uses TMDB data but is not endorsed or certified by TMDB.

Indx Search Platform

A search platform with AI on both sides

An open-source server for your datasets and a console to configure them and see how visitors search. Its AI works on both sides: it answers your visitors’ questions from your own records, and tells you what to tune.

Get it on GitHub Read the docs

See what your users are doingTune relevance without codeAnswers questions, not just searchesAI that tells you what to tune

The Indx Server console on the Field configuration page of a demo furniture catalogue.

The real console, on a demo furniture catalogue.

Searched, chosen, counted

Search statistics

See what visitors search for, what they choose, and where they choose it. Counted by your own server, with nothing sent anywhere else.

Learn more →

Top queriesFuzzy findsClick-through

The console's Statistics page: searches, fuzzy finds, click-through and click position over 30 days, as a chart.

The most searched queries, with click-through and average position.

Features

Chosen, lifted, scheduled

Boost rules

Lift the products you want seen first: top rated, best sellers, this season's campaign. Build a rule from your own fields, give it a strength, and let it start and stop on the dates you choose.

Learn more →

Your own fieldsStrength per ruleScheduled campaigns

The console's Boost rules: a Halloween campaign running now, and a Christmas one that starts on 20 November, each built from conditions on the catalogue's fields.

A top rated rule: rating from 4.5 and at least 100 ratings.

Features

Typed, judged, answered

Search, or ask

Visitors search as they type. When what they type reads like a question, the list already searches what it is about, and they are offered an answer written only from your own records, with the products it used.

Judged while typingSearches what it is aboutAnswers from your records

The console's Search or ask tab: its settings, and the tester beside them, where "Can you help me find a velvet sofa?" searches for velvet sofa and offers an answer.

When to offer: the score where the quiet invitation and the full offer to answer start, on a scale from lookup to question.

Features

Watched, flagged, proposed

Observr

An AI that watches your search. Each morning it reads your datasets' statistics, flags what needs a look, and proposes synonyms, boost rules and field changes. Nothing changes until you say yes.

In Indx Professional →

Every morningDiscuss it with the agentWaits for your yes

Observations on a dataset's Status tab: sizes typed differently, other shops' names in searches, a query rarely chosen, and an answer about the field setup.

Suggested changes: make description searchable with weight 0.5, and pattern with weight 1, each with its reason.

Features

Indx Search Library 5.0

A full enterprise search in a library

The same engine as a NuGet package for .NET: search that recognises patterns in text instead of matching exact words, with filters, facets, boosts and vector search, in-process and with no server to run.

One package, in your own .NET code

Get it on NuGet Download example projects

Pattern-based matching

Returns relevant search results automatically, even with messy or distorted input.

Great for AI agents

Fast structured retrieval for agent workflows and hybrid RAG pipelines.

Native C# engine

A fully embeddable search engine with no dependencies.

Zero tuning

No tokenizers, analyzers, or configs. Simple levers when you want control.

Fast as visitors type

Results in milliseconds, on every keystroke, in your own process.

Made for catalogues

Facets, filters and boosts that help visitors find the right product, and lift what you want seen.

Split-word understanding

Handles combined and separated word forms automatically.

Ready for production

Real-time updates, reindexing without downtime, and filters that hold up under load.

Index and search in a few lines of C#

Point Indx at your JSON, mark a few fields as searchable, and run your app.

csharp
using Indx.Api;
var engine = new SearchEngine();

// ANALYZE AND INDEX JSON
FileStream fstream = File.Open("movies.json", FileMode.Open, FileAccess.Read);
engine.Init(fstream);

engine.GetField("title")!.Searchable = true;
engine.GetField("description")!.Searchable = true;

engine.Load(fstream);
engine.Index();

// SEARCH
var result = engine.Search(new Query("matrix", 10));