AnswerThis reads 250 million papers so researchers don't have to
Two University of Richmond seniors built an AI research assistant that turns a month of literature review into an afternoon. Y Combinator noticed. So did a quarter-million researchers.
There is a moment every graduate student knows. You have a research question, a deadline, and somewhere out there sit five million papers published this year alone, any handful of which might already answer it - or prove you wrong. The literature review is the tax you pay before the interesting work begins, and for most researchers it costs weeks. AnswerThis was built to shrink that bill to an afternoon.
The company, part of Y Combinator's Fall 2025 batch, is an AI research assistant that searches a database of more than 250 million academic papers and answers complex scientific questions with citations attached to every line. Ask it something specific - the effect of a compound, the consensus on a contested finding - and it returns a synthesized answer where each sentence links back to the paper it came from. You can click, verify, and save the source without leaving the page.
The problemResearch runs on reading, and no one can read fast enough
Roughly five million research papers are published every year. A diligent human reads a few hundred. That gap - between what is known and what any single person can hold in their head - is where AnswerThis lives. The literature review is repetitive, slow, and it pulls scientists away from the part of the job that is actually science: the experiment, the argument, the discovery.
The founders like to point out that this is not a niche annoyance. Every professor, every PhD candidate, every lab tech doing a first-pass scan of a field runs the same errand, over and over, in every discipline. AnswerThis treats that errand as the market.
How it worksSearch, read, synthesize, cite
The workflow is deliberately linear, mirroring how researchers already move from question to draft.
Ask
Pose a scientific question in plain language across 250M+ papers.
Verify
Every answer arrives with line-by-line citations you can open.
Map
Citation maps surface foundational and highly-cited work on the topic.
Write
Draft in an integrated canvas; export with citations and equations intact.
The differenceNot another chatbot that invents its sources
General-purpose AI tools have a well-documented habit: ask for a citation and they will sometimes hand you a plausible-looking reference to a paper that does not exist. In most contexts that is embarrassing. In academic research it is disqualifying. AnswerThis made the opposite bet from the start - the claim-to-source link is preserved at the sentence level, so every specific assertion traces to its originating paper.
That focus shows up in features general tools don't bother with: a Research Gap Finder that reads existing literature and points to unexplored areas, contradictions, and open questions; citation maps that visualize how a field cites itself; and exports that keep equations and references clean. Integrations with Zotero and Mendeley mean the citations flow into the tools researchers already keep their libraries in.
Scope a new topic in an afternoon. Draft a citation-backed literature review in one tab. Find the gap nobody has studied before you commit a year to a thesis. Trace any claim to its source without a dozen browser tabs.
The foundersThe son of two PhDs, and a coffee entrepreneur
Ayush Garg, the CEO, is the son of two PhD scholars. He published his first peer-reviewed paper at 16 - on software that mines and parses scientific literature - and did early research with the University of Cambridge's semanticClimate group on text and data mining. The users AnswerThis serves are, in a real sense, the people he grew up around.
His co-founder and roommate, Ryan McCarroll, runs product and growth. Before this he founded Crackin' Coffee, a subscription coffee startup that shipped internationally, and Deep Point Lab, a web and marketing shop. The two met at the University of Richmond - Garg studying computer science, McCarroll mathematical economics - and built AnswerThis into the first company founded there to be accepted by Y Combinator, selected from roughly 10,000 applicants at about a one percent acceptance rate.
The businessFreemium, aimed at the people who feel the pain
AnswerThis sells the way most research software does: a free tier with monthly AI credits and storage, then paid Plus (around $12 a month) and Pro (around $24 a month, billed annually) subscriptions that unlock deeper reviews and more credits. It goes direct to individual researchers - professors, doctoral students, lab members - with a natural path toward teams and institutions. The founders have said the reported figure is over $1M ARR within eight months of launch.
It sits in a crowded and fast-moving corner of AI, alongside Elicit, Consensus, SciSpace, Scopus AI and the general tools researchers reach for out of habit. AnswerThis's wager is that specialization wins: an all-in-one workspace built for citation accuracy beats a clever assistant that treats a research question like any other prompt.
The ambitionLiterature review is only the first errand
The stated mission runs bigger than papers. The founders frame AnswerThis as an end-to-end workspace for scientific discovery, with an eye toward eventually helping any knowledge worker across any knowledge format. The framing is unapologetically grand - accelerating science, extending healthy human lifespans, a multiplanetary future. For now, the concrete version is narrower and more useful: give researchers back the weeks they currently spend reading.