NeoSearch opens source code to boost search transparency

NeoSearch has released its search engine code under the Apache Licence 2.0, giving the public access to inspect and run the software.

The Sydney-based search company says users, developers and researchers can review how its system handles queries, applies privacy measures and re-ranks search results. Others can also fork the code and run their own version of the service.

NeoSearch presents the decision as part of a broader push to make search more transparent in a market where major platforms keep their systems closed. It says the service does not log IP addresses, set cookies, build user profiles or retain search history.

Founder Bart Jellema said the release is meant to allow outside scrutiny rather than ask users to accept privacy claims at face value.

"Privacy claims mean little if you have to take them entirely on trust," said Bart Jellema, Founder, NeoSearch. "Opening our code lets anyone see how NeoSearch works, challenge it, improve it or run their own copy. This isn't a product built to be sold or to harvest data. It's meant to be a lasting public utility, and open source ensures the community always has the tools to carry it forward."

Results model

Alongside the code release, NeoSearch outlined how its search product differs from conventional results pages. It says it uses artificial intelligence to assess whether results are commercial, official or independent, then adjusts rankings to favour what it considers authoritative and independent sources while pushing down search engine optimisation-driven pages such as affiliate list articles.

It also rewrites titles and snippets so users see shorter descriptions without keyword-heavy language. Results on the same subject are grouped into a single entry, and the interface sorts findings into what it calls Lenses and categories, with optional AI summaries and instant answers for common searches.

Jellema said these choices respond to the commercial pressures that shape many search rankings.

"Search results today are shaped by whoever spends the most on SEO, not by who has the best answer," Jellema said. "We built our ranking to push back against that. Independent, authentic sites deserve to be found, and users deserve results chosen for relevance rather than commercial muscle."

Staged approach

For now, NeoSearch relies on top results from Google's index as its starting point. It then reorganises, re-ranks, rewrites and enriches those results. According to NeoSearch, that process can lift a result that appears much lower on Google to the top position on NeoSearch.

The approach reflects a staged development model. NeoSearch says the current setup is intended to show that a different search experience can be built on top of existing search data before it expands its own infrastructure.

The next step is to add its own web crawling, focused on parts of the internet that attract less attention from larger search engines. Longer term, NeoSearch plans to create an open distributed search index that others can use as a base for their own services.

Jellema said the aim is not to build another dominant search platform but to widen access to search infrastructure.

"The goal was never to become the next search behemoth," Jellema said. "The goal is to liberate search. If a thousand different search engines one day run on top of an open index, each with its own ideas about ranking and design, the mission is complete."

Funding model

NeoSearch says the service is free to use, carries no advertising and does not profile users.

Jellema is personally funding operating costs. His previous work has included 3D graphics, database systems and the co-founding of coupon aggregator Tjoos, which was later acquired. For longer-term revenue, NeoSearch expects to rely on developer API access to curated data and on disclosed affiliate links that do not affect rankings.

The code release places NeoSearch among a small group of search projects making their software publicly available at a time of continued scrutiny over ranking systems, privacy practices and dependence on a handful of dominant indexes. Its immediate proposition is less about replacing the underlying web index than showing how a search interface can sort, present and prioritise results differently.

Anyone can now examine the software, test its privacy design and deploy an independent instance under the Apache Licence 2.0.