Parsem is an AI-powered search tool that makes sources visible again. It highlights original reporting, makes AI-derived content verifiable in-app, and lets you navigate directly to the material informing the AI’s answer.

AI has transformed how people find information. This shift has consequences.

When users stop clicking through to original sites, publishers lose traffic, depleting the funding for new, factual, human-generated work. Parsem is designed to address this problem by giving users clear, intuitive paths back to the source. Wherever possible, it embeds the source directly beside the AI’s text, helping to reverse the dramatic decline in referral traffic that threatens independent media.

Every piece of writing has a perspective. Today’s AI systems blur those perspectives until the origins of information disappear entirely. When a model answers based only on its training data, not even the AI knows where its words come from — the bias is hidden. When it uses retrieval-augmented generation (RAG), multiple sources are collapsed into a single opaque citation. Parsem is built differently. Each section of the AI’s response is tied to one specific source, with a direct citation and visible quote. Rather than erasing a writer’s viewpoint, it surfaces it, letting readers see how ideas and evidence are connected — and form their own judgments.

This project began as an essay about journalism’s crisis in the age of AI — how answer engines like Google AI and Perplexity are reshaping the economics of the web and the public’s relationship to truth. At its core, Parsem asks how AI and human knowledge can coexist. Instead of collapsing the web into a single, authoritative answer, it makes AI a guide — one that still points you back to the people who did the work.

Interested in Parsem or in collaborating? Visit alexduncangames.com for contact information.

© 2025 Alex Duncan.