AI Writer
Answers research questions from 100 million open science papers, with every claim traceable
Answers research questions from 100 million open science papers, with every claim traceable
Fast-read signals for fit, pricing, and trust.
The reason to use this rather than a general assistant is verifiability: claims trace to specific papers and paragraphs rather than to a model's memory, which is what makes the output usable in work that will be checked. Two things to establish before subscribing: the plans hide their question allowances behind an unexplained asterisk, and the jump from Standard at $49 a month to Power at $375 is steep enough that you need to know what you get for it.
The name is the most misleading thing about this product. It is not a copywriting assistant; it is a research tool that answers questions from published science and shows where each part of the answer came from. Ask it something and it returns a summary, a detailed breakdown, and citations that resolve not to a paper but to the paragraphs inside it. For anyone writing something that will be read critically, that granularity is the whole point — a citation you can open and check in one click is a different kind of evidence from a plausible-sounding sentence.
What sits underneath is a collection of more than a hundred million open access papers, and that choice defines both what it is good at and where it stops. In fields that publish openly it has genuine depth. In fields where the important work is paywalled, it will answer confidently from what it can reach, which is a narrower slice than the field itself. Topic Explorer is the other half of the offer and the more unusual one: given a subject, it proposes the questions a comprehensive treatment would need to answer, which is the part of a literature review that is hardest to do from outside a field.
The commercial picture is straightforward at the bottom and opaque in two places. Basic is $29 a month for one user and Standard $49 for three, both with annual rates that save about two months. Power is $375 a month for ten users — a step of more than seven times for three times the seats, which implies something else changes at that tier without saying what. The plan cards also mark their question allowances with an asterisk whose footnote is not published, so the one number that would tell you which plan you need is missing. Against that, the trial is unusually clean: seven days, no card, no auto-renewal, and it ends by itself.
Quick fit check against how you actually work.
What you can actually do with this tool.
Citations land on the specific passages the answer drew from, one click away, with a direct link and full bibliographic information. Check a claim and you land on the sentence it came from, not on a reading list.
The reading it draws on is open access literature rather than the open web, which sets both the strength and the boundary: strong where a field publishes openly, weaker where the important work sits behind paywalls.
Given a subject, it works out which questions a comprehensive treatment would have to answer. For someone entering an unfamiliar literature, knowing what to ask is the harder half of the problem.
The same engine handles a single question and a structured review, so an exploratory session can turn into a document without starting again in another tool.
Seven days, no credit card, and the vendor states it does not become a subscription and does not need cancelling. That removes the usual reason to hesitate, and it makes trying it the cheapest way to answer the questions the pricing page leaves open.
Pricing tiers and what's included in each.
Common questions about this tool, answered.
Provenance. The answers are generated from a corpus of published open access papers, and every claim links back to the specific paragraphs used, with a direct link and full bibliographic detail for each source. A general assistant can produce a similar-sounding paragraph and a plausible citation list; this shows you which passage each statement rests on. If nobody will check your work, that difference costs you money for nothing. If someone will, it is the entire reason to use it.
Not published. The plan cards carry an asterisk against the question limit and the footnote explaining it is not on the page, and there is no separate pricing page carrying it. The visible difference between the tiers is seats — one, three and ten. Since that is the number most likely to determine which plan suits you, establish it during the free trial or with the vendor before subscribing.
Nothing, on the vendor's own terms. It runs seven days, requires no credit card, does not turn into a subscription and ends automatically without being cancelled. That is a stronger statement than most trials make, and given how much of the pricing detail is unpublished, it is also the practical way to find out what a plan would give you.