About Bionoculars
Researchers today face a real dilemma.
Traditional search engines like PubMed or Google Scholar are exhaustive, but they demand enormous manual effort. AI tools are fast, but opaque, selective, and prone to hidden biases that quietly undermine genuine exploration.
Bionoculars was built to resolve that tension.
It's a search engine for biomedical literature, covering medicine, biology, biochemistry, microbiology, and more, designed to hit the sweet spot between classical search and AI-assisted research. A tool that uses AI to serve transparency, control, and intellectual sovereignty, not to replace them.
Our ambition
Build a transparent, exhaustive AI-assisted research tool that empowers scientific reasoning rather than automating it away.
The problem
The design of Bionoculars is driven by specific observations about how research tools work today.
A growing body of research suggests AI is starting to narrow science. Recent studies indicate that while AI boosts individual productivity, it may reduce the collective diversity of topics studied and diminish researcher-to-researcher engagement (Hao et al., 2026; Messeri & Crockett, 2024). We designed Bionoculars so the user steers exploration, not the algorithm.
Experts use different words to mean the same thing. A gene like "prion protein" can appear as "PDCD1", "Programmed cell death protein 1" or "CD279"depending on the author and the context. Keyword groups handle this mapping systematically, so researchers can focus on their actual question instead of guessing which synonym the index uses.
Search intent is difficult to express upfront, and even state-of-the-art AI can't guess it. When exploring a new topic, researchers rarely have the perfect query to express their needs from the start. That's why we designed our AI so it can be introduced at any step of the search process, rather than trying to interpret intent from a single prompt. Keyword groups make the search logic visible and editable, so intent can develop iteratively.
Current tools force a false choice. PubMed gives you everything but little help navigating it. AI chatbots give you a shortcut but strip away your ability to verify and explore. We wanted both depth and assistance.
Data sovereignty matters. Three American tech giants control more than 70% of Europe's cloud infrastructure, and the question of where research data lives is gaining regulatory attention. We host exclusively on European providers (OVH, Netcup) and do not sell user data.
Our approach
Every feature in Bionoculars is designed around four ideas:
Transparency
Every AI-generated insight links back to its source articles. The data used to rank results, including semantic keyword groups, is shown directly, not hidden in a model.
Control
Researchers steer their own exploration. Queries can be refined, sources resorted, and reasoning built iteratively. AI operates on the user's selection, never on a black box.
Exploration
Semantic keyword groups make the ranking logic visible and editable. A knowledge graph maps real conceptual relations, each link backed by actual articles. New research directions emerge from evidence, not from suggestions.
Efficiency
AI actions help synthesize and prioritize across a selection of articles. Not to produce a summary to trust blindly, but to help move faster from a large corpus to focused reading, without losing the thread back to the source.
What it does
These translate into concrete features:
- Keyword groups: Results are indexed using semantic keyword groups built on UMLS Metathesaurus data. You can pin, exclude, reorder, and merge them to reshape and explore results before running a new search.
- Knowledge graph: Maps relationships between keywords extracted from articles. Each link is backed by actual papers, allowing further exploration of the research subject.
- AI actions: AI operates on your article selection and order. Every claim in a summary is cited back to source articles. You choose what the AI reads.
- Searches & Projects: Searches are saved automatically (query, keywords, filters, selections). Projects let you collect articles across searches and run AI overviews on the combined set.
See it in action
Example: Calcium score in nuclear medicine. A walkthrough showing how keyword groups helped a researcher surface relevant papers that other tools buried in pages of less relevant results.
More examples coming soon.
Tutorial. Step-by-step guide covering search, exploration, AI actions, and projects.
Who we are
We are a small, independent team based in France, building AI tools that help researchers work more effectively, tools that serve transparency and scientific reasoning, not replace them.
An honest note
The product is in early access. There are rough edges: the article index is incomplete for some closed-access journals, and we ship improvements continuously. Early users directly shape what Bionoculars becomes.
If something feels off, missing, or could work better for your workflow, contact us. We take every piece of feedback seriously.