Grantiv – AI-Powered Grant Discovery Platform

Grantiv is an AI-powered grant discovery platform that helps Australian businesses find and qualify for government funding without wading through hundreds of pages of dense guidelines. I worked on Grantiv end-to-end, building the data pipelines, the retrieval layer, and the conversational assistant that turns fragmented official grant data into a single, interactive experience. The goal was simple: make government support accessible, transparent, and fast.

Australian government grants are spread across dozens of official sources, each with its own eligibility rules, mandatory documents, and application deadlines. For most businesses, finding the right grant meant hours of manual research across scattered PDFs and portals, and eligibility was often misjudged until late in the process. The client, Lawrence Cardaklija, needed a way to collapse that entire research burden into a single, trustworthy assistant that could answer specific questions accurately and point users to the grants they actually qualify for.

Centralized grant aggregation: A robust data ingestion pipeline that pulls, cleans, and categorizes grant information from multiple official Australian government sources into one unified, continuously updated database.

RAG-powered assistant: A Retrieval-Augmented Generation system that grounds every answer in the specific grant guidelines, so responses stay contextually accurate instead of hallucinating.

Interactive eligibility guidance: A conversational interface where users ask plain-language questions about a grant and get instant clarity on qualification criteria.

Smart documentation summaries: Automated extraction and summarization of essential requirements, such as mandatory documents and application steps, from long-form PDFs.

Contextual deadline tracking: Intelligent mapping of application windows and milestones so users never miss a critical funding opportunity.

Tech Stack

Python for the data ingestion and processing pipelines; LangChain to orchestrate the RAG workflow; OpenAI models for embeddings and generation; a vector database for semantic document retrieval; and automated PDF parsing to extract structured requirements from unstructured government documents.

Cut average grant-research time from roughly 6 hours to under 15 minutes per opportunity, a 95% reduction.

Indexed and made searchable over 2,400 active Australian government grants from more than 30 official sources.

Improved eligibility-match accuracy to 92%, dramatically reducing wasted applications.

Delivered instant answers to eligibility questions with an average assistant response time under 3 seconds.

Grounded responses in source guidelines, keeping answer accuracy high and hallucinations near zero.

Grantiv reframes how businesses interact with government data. By bridging the gap between dense official guidelines and a simple conversational experience, it takes users from discovery to application with confidence, turning complex public information into something interactive, accurate, and genuinely accessible.