An AI tool that turns a scientific paper PDF into an interactive, easy-to-understand web page. Users upload an English-language paper, choose a mode (Native technical, Non-technical, or Kid-friendly), and the system uses Claude to generate a standalone narrated page in about 30–60 seconds—including analogy boxes, data-visualization cards, and clickable expandable details—rather than a flat summary. Each result has its own shareable link. It is operated by John Damask (AWS background) under Amroja LLC (Somerville, MA), built on AWS (Lambda + S3 + DynamoDB) with Claude Opus 4.6 for generation and Haiku for input safety screening. Pricing is credit-based pay-per-use with no subscription and non-expiring credits (Future Tools marks it Paid); an earlier free allowance of 20 papers/day across the site was the author\'s self-funded phase and has since shifted to paid. Critical caveats: It supports English papers only; generated content depends on the LLM and can contain inaccuracies—one Hacker News user reported charts in the output that were not in the original paper, which the author acknowledged as “conceptual illustrations” rather than paper data, and another user found the explanation of Query/Key/Value in “Attention is All You Need” to be incorrect; ScamAdviser gives it a low trust score of 55/100 (WHOIS privacy, very new domain registered in February 2026, high-spam registrar), though SSL is valid and DNSFilter marks it safe; and no independent third-party reviews (G2/Capterra/Trustpilot) surfaced.
Pros & Cons:
Pros: The output is a structured interactive web page with analogies, visualization cards, and expandable detail, rather than a flat text summary—this is a genuinely different format from most paper-summarization tools; three audience modes (technical, non-technical, kid-friendly) let the same paper be explained at different levels; results are shareable via a standalone link, which is useful for lab websites, teaching, or team reading groups; credit-based pay-per-use with non-expiring credits suits occasional users better than a monthly subscription; the author is publicly transparent about the product\'s limits and iterates in the open on Hacker News (expanded context window to support papers of hundreds of pages, added modes).
Cons: The core risk is LLM hallucination in a tool whose purpose is comprehension—HN users reported charts not present in the original paper and an incorrect explanation of a key concept, and generated speculative content is not always clearly labeled as AI inference, so non-expert readers may mistake it for the paper\'s own conclusions; independent validation is essentially absent—no G2/Capterra/Trustpilot reviews, and ScamAdviser\'s 55/100 reflects a very new domain and hidden WHOIS; it supports English papers only, which limits its use for non-English research; it is a “comprehension aid,” not an authoritative reading—it cannot replace reading the original paper; and there are emerging alternatives with more institutional backing (e.g., academic systems like Paper2Web/PWAgent, and Google\'s NotebookLM for document explanation), so its differentiation depends on staying ahead on quality.
Who is it for?:Researchers and students who need to quickly grasp a paper outside their own field before deciding whether to read it in depth; lab teams and reading groups that want a shareable, structured explainer page for a paper; teachers and science communicators who need to present a paper at different audience levels; and anyone who prefers an interactive web format over a plain text summary. Not suited for users who need an authoritative or citation-grade interpretation of a paper, non-English papers, organizations requiring independently validated accuracy, or anyone who needs a mature product with substantial third-party reviews.