Draft:Indexa

InBot and early development

The earliest component of what would become the Indexa platform was a web crawler originally developed to monitor and track pricing data for PC components. At this stage the crawler operated under the name PCBOT, and was focused narrowly on scraping and aggregating hardware pricing information from across the web.

PCBOT was later repurposed and expanded to serve as the data backbone for an unreleased AI product developed internally at AWNI Tech. This product made use of Meta AI's open-source LLaMA language model and represented AWNI Tech's first venture into AI-driven applications. Though the product was never publicly released, the development work established the infrastructure and experience that would feed directly into Indexa.

The crawler was subsequently generalised for broad web crawling — indexing pages across a wide range of domains rather than targeting specific product categories — and was renamed InBot. With this expansion in scope, InBot became the foundation on which the Indexa search engine was built.

Indexa

Indexa was created once InBot had matured into a general-purpose crawler capable of building and maintaining a searchable index of the web. The platform launched with support across three form factors: a Mobile and Tablet version, a Desktop version, and a TV version, reflecting an intent to make Indexa accessible across the range of devices through which Indian users access the internet.

Indexa Spaces

Indexa Spaces originated as an internal tool built by Yohaan D'souza to assist with personal study and research — a conversational AI assistant that could surface and contextualise information from Indexa's index. Recognising the broader utility of the tool beyond its original personal use, D'souza developed it into a standalone product and released it publicly as Indexa Spaces.

Dufus

Dufus is AWNI Tech's in-house machine learning model, currently in early development. Unlike a general-purpose language model, Dufus is oriented specifically toward understanding human behaviour as it relates to information retrieval — modelling how people think, how they formulate queries, and how their search behaviour reflects underlying intent. The goal is to build a model that can interpret and respond to natural, imprecise human queries more accurately than systems trained on text generation alone, with comprehensive text generation as a secondary capability built on top of this behavioural understanding.

Mission

A founding principle of AWNI Tech and the Indexa platform is that access to a high-quality, independent search engine should carry no cost to the user and no compromise on privacy. Indexa does not collect user data, does not serve advertising, and is built with the explicit goal of keeping the internet free and accessible. The platform is designed to be safe, secure, and private by default — with no data collection as a structural commitment rather than a policy preference.

Indexa Spaces

Indexa Spaces is an AI-powered learning and discovery application developed by AWNI Tech. Rather than returning a list of links, Indexa Spaces allows users to explore topics through a conversational and structured interface, drawing on Indexa's search index and knowledge graph to surface relevant content. The application uses the Dufus language model to generate contextual answers, topic summaries, and suggested learning paths based on a user's query.

Indexa Spaces is designed to support extended engagement with a subject — allowing users to progressively explore related concepts, save and organise information, and receive curated content derived from the broader Indexa index.

Dufus

Dufus is AWNI Tech's proprietary language model, built to power the AI features of the Indexa platform. It handles natural language understanding of search queries, generates text summaries for Indexa Spaces, and assists in populating Indexa's knowledge graph with structured entity data extracted from crawled pages.

Dufus is tightly integrated with InBot and Indexa Spaces. Query inputs from Indexa Spaces are processed by Dufus to produce contextual responses, while structured data surfaced by InBot's crawls is used to keep Dufus's knowledge base current.

InBot

InBot is AWNI Tech's web crawler and knowledge graph generation system. It systematically crawls and indexes web pages to build and maintain Indexa's search index, and simultaneously extracts structured information to generate a knowledge graph — a database of entities such as people, organisations, places, and concepts, along with the relationships between them.

This dual function means InBot goes beyond standard page indexing: as it crawls, it identifies and links named entities, enabling Indexa's search results to surface structured information panels and enabling Dufus to give more accurate, entity-aware responses within Indexa Spaces.^[1]