QED Science launches AI-powered Insights Feed to help researchers find quality papers

Why researchers need better discovery tools

According to The Next Web, researchers today struggle less with finding scientific papers and more with deciding which ones are worth reading. The volume of research has grown significantly, with major databases indexing roughly 47% more articles in 2022 than in 2016. Generic AI tools often exacerbate the problem by generating longer reading lists without improving judgment, shifting the bottleneck from discovery to evaluation.

The decline of platforms like Twitter, which once served as a hub for scientific discussion and discovery, has further complicated this issue. A 2025 survey by David Shiffman and Julia Wester found that two-thirds of scientists believed Twitter had become much less useful for learning about new developments in their fields. This gap has left researchers seeking alternative ways to curate and evaluate scientific literature effectively.

How Insights Feed works

Insights Feed, launched by QED Science, aims to solve these challenges by providing a personalized discovery experience for life-science researchers. The tool builds profiles based on researchers’ published work, stated interests, and uploaded manuscripts or proposals. It then recommends papers, preprints, and conference abstracts tailored to their specific needs.

One of the tool’s standout features is its ability to flag papers for relevant methods or results, even if they come from unrelated fields. This expands the scope of discovery beyond traditional search parameters. Preprints are a particular focus, as they often contain findings months or years before formal journal publication.

To help researchers assess the quality of preprints, Insights Feed includes the QED Score, a metric derived from QED Science’s scientific judgment engine. This score provides a signal of scientific rigor, addressing a key limitation of generic AI summaries, which often lack the ability to distinguish between high- and low-quality studies.

The platform is free for academics and covers 250 life-science subfields, reflecting its goal of making high-quality research more accessible.

Adoption and early traction

QED Science’s scientific judgment engine is already used by more than 20,000 researchers across over 70 countries. Insights Feed builds on this existing user base, offering a more streamlined and personalized way to discover relevant research. By integrating quality signals like the QED Score, the tool aims to recreate some of the curation and evaluation functions that platforms like Twitter once provided.

What this means

Lazyfounder analysis — our interpretation, not reported fact.

Insights Feed reflects a broader shift in how researchers interact with scientific literature. The challenge is no longer about finding papers—it’s about filtering them for quality and relevance in an era of information overload. Generic AI tools have struggled to address this, often adding to the noise rather than cutting through it. QED Science’s approach, which combines personalization with quality signals like the QED Score, could set a new standard for how researchers discover and evaluate preprints and published work.

For founders building tools in this space, the takeaway is clear: personalization and quality assessment are critical differentiators. The decline of science-focused social media has left a void, and startups that can recreate the curation and discussion once enabled by those platforms will have a clear edge. Preprints, in particular, represent a massive opportunity, as they often contain groundbreaking findings long before they appear in journals. Tools that can effectively surface and evaluate them will fill a critical gap in the research ecosystem.

Key takeaways

• For researchers, Insights Feed shifts the focus from finding papers to evaluating their quality, addressing a key pain point in scientific literature discovery.
• Founders in AI-driven research tools should prioritize quality signals like the QED Score to differentiate their products in a crowded market.
• Preprints are a critical source of early findings, and tools that effectively surface and evaluate them can fill a gap left by traditional discovery methods.
• Personalization is key to improving researcher productivity, especially as generic AI summaries fail to distinguish between high- and low-quality studies.
• The decline of science-focused social media platforms like Twitter highlights an opportunity for specialized tools to recreate curated, high-signal research communities.

Related on Lazyfounder

• More technology news