Stop writing Boolean queries — just ask what you want to know

"crispr AND off-target AND mouse NOT review" — if you've ever typed something like this into PubMed, you know the pain. Miss one synonym and you lose papers. Add the wrong term and you get buried in noise. It's 2026 and we're still writing queries like it's 1998.

That made sense when databases could only match exact strings. It doesn't make sense now. Language models understand meaning. So we built a search that understands meaning too.

On NobleBlocks, you just type what you'd say to a colleague:

• "What's the latest evidence that GLP-1 agonists protect the kidney?"
• "Randomised trials comparing ketamine to ECT for treatment-resistant depression"
• "Has anyone replicated the 2018 Stanford dermatology diagnostic model?"
• "CRISPR delivery mechanisms in vivo, published after 2023"
• "Long COVID neurological symptoms — recent reviews only"

We figure out what you actually want — study type, population, intervention, outcome, date range — and return papers that match on meaning, not just keywords. Ask about "kidney protection" and we'll also find papers about nephroprotection, renal outcomes, and eGFR decline. You don't need to list every synonym yourself.

What's happening under the hood

Three systems working together: semantic embeddings that capture meaning, lexical matching that catches exact terms like gene names and drug codes, and citation signals that help surface influential work. Plus study-design filters so you can narrow to RCTs or meta-analyses without adding extra query terms.

You don't need to know any of that to use it. Just type your question and see what comes back. It's usually faster than writing a Boolean query, and it finds papers you'd miss with keywords alone.