ΕΡΓΟΝ Β΄
Knowledge IntelligenceAPI Only, No App YetCiteWeave
“Understanding a field shouldn't require fifty browser tabs.”
CiteWeave is a research tool. Submit a paper, and it shows you where that paper actually sits in its field: what it builds on, what disagrees with it, what's genuinely new about it. It does this today by following the real trail of citations between papers and pulling real gap-revealing sentences straight out of them. The one thing missing is a web page to use it from, that's still ahead.
Timeline
Loading diagram…
Ι — Origin
How it started.
CiteWeave came out of research frustration. As an AI researcher, finding information isn't difficult. Understanding it is. Researchers keep asking the same questions. What datasets get used? What metrics matter? What's actually novel here, and what's foundational? Existing tools retrieve information. Orientation still costs you a week.
ΙΙ — The Problem
What we're solving.
Most AI search tools are lazy. They take your question, find text that shares similar keywords, and have an AI write an answer from that. In academia, law, and medicine, that's fatal. It ignores how important a paper actually is: it will treat a 1990 abandoned hypothesis the same as a 2024 breakthrough with a thousand citations, simply because the keywords matched.
People don't want more papers. They want to know how a paper fits: what it's next to, what it disagrees with, what it actually explains. Literature should feel explorable, not exhausting.
ΣΦΑΛΜΑ — What We Got Wrong
The first version was a single hardcoded pipeline built for academic papers only. Adding a second domain meant forking the whole codebase, and every future domain would need its own fork forever. We rebuilt the core to be domain-agnostic, then proved it by running legal case-law retrieval on the same engine, not a copy of it.
ΙΙΙ — Where It Stands
Told straight, no jargon.
CiteWeave answers the question every researcher asks at one in the morning: where does this paper actually sit in the field? Give it a paper, and instead of a list of similar-sounding results, it hands you a map. What it builds on, what disagrees with it, what's genuinely new about it.
The engine behind that map is real and running. It doesn't just match keywords. It follows actual citation trails, and it pulls real sentences out of papers' "what we couldn't figure out" sections instead of guessing at gaps.
Today it's an engine, not an app. There's no web interface yet, you'd need to be a developer to query it directly. That's the next thing to build, and we'd rather build the right interface than the fast one.
Want the wiring diagrams, the real stack, and what’s still 0% built?
IV — Today
What it can already do.
- ›Follows which papers actually cite which, weighing real influence over keyword overlap
- ›Finds research gaps from real sentences pulled out of papers, not AI guesses
- ›Runs its own AI model in-house, no dependency on an outside company's service
- ›The same engine works for law or medicine with a settings change, not a rewrite
V — Who It's For
Who we're building for.
PhD Students & Academic Researchers
"Query it directly today, as a developer, and get context verified against the real citation trail. The engine is real, even before there's a browser tab to put it in."
R&D Labs & Big Tech
Surfaces the benchmarks, datasets, and metrics that matter in your domain, automatically, integrable into your own tooling today.
Law Students & Paralegals
"The retrieval pipeline for case law already exists and is tested. We just haven't exposed it yet. Say the word and it jumps ahead of everything else on the roadmap."
ΜΕΣΑ — Media
CiteWeave demo, a literature-review query end to end
VII — Open Questions
What we don't know yet.
- Is it right to launch for developers first, or does this need a real screen to type into before anyone actually uses it?
- Can machine-generated orientation ever become something academia trusts?
- What domains need this most urgently, beyond CS research and law?
- How much synthesis is useful before people stop verifying it themselves?
Feedback wanted from: Researchers, academics, lawyers, scientists, policy professionals, anyone doing literature reviews or fed up with existing discovery tools.
We’re not asking you to validate this. We’re asking you to break it.
ΟΜΑΔΑ — Team
Nikhil Y N
CEO & Technical Lead
Praneel S
Engineer / Manager, Frontend & SEO
Karan R A
Engineer, Python Backend & AI/ML
Manichandan M
Engineer, Java & Web Development