Research Assistant

Match your draft to the right journals and find prior work to cite — fully on your device. No uploads, no account, free.

Research Assistant helps graduate students and early-career researchers answer two questions every author faces: "Where should I publish this?" and "What should I be citing?" — without ever sending your manuscript to a server. Everything runs on your own device. There is no backend, no account, and no tracking. Your draft text never leaves your laptop; only short, anonymous phrase queries are sent to public research databases, exactly like typing a search into PubMed. ★ THREE TOOLS, ONE SIDE PANEL 1. Where could I publish this? (Journal Matcher) Paste your abstract and introduction. An AI sentence-embedding model runs locally in your browser and ranks a curated set of journals by how well your draft fits their scope — with similarity scores, open-access status, approximate article-processing charges, decision times, and a direct link to each journal's submission page. 2. What should I cite? (Prior-Art Finder) Paste a section of your draft. Research Assistant samples distinctive phrases and searches PubMed, OpenAlex and Crossref in parallel for prior work that resembles your writing — so you can find the literature you should be citing. Filter by source and year, sort by relevance or recency, and export to BibTeX, RIS or CSV. 3. My drafts library (Self-overlap check) Keep a private, on-device library of your own earlier papers, preprints and theses. When you check a new draft, Research Assistant highlights passages that overlap with your own prior work — useful for thesis writers reusing conference-paper material who need to disclose or cite it correctly. This runs fully offline. ★ PRIVATE BY DESIGN, NOT BY PROMISE - Your manuscript is processed entirely in your browser. - There is no server that holds your drafts — the architecture makes collection impossible. - No analytics, no telemetry, no advertising, no account. - A built-in "Verify privacy" panel shows you every network request the extension makes and the short allow-list of domains it may contact. You can confirm it yourself in DevTools. ★ FREE — AND IT STAYS FREE Every feature is available to everyone. There are no subscriptions, no paid tiers, and no locked features. Email is entirely optional and opt-in (you can leave it blank forever). ★ WHAT THIS IS NOT Research Assistant is a discovery and self-check assistant — not a plagiarism checker and not an AI-content detector. It surfaces similarity and prior work to help you improve and cite correctly; it never accuses your writing of plagiarism, and it contains no AI-text "detector" (those have high false-positive rates against non-native English writers, and we will not ship one). ★ HOW IT WORKS UNDER THE HOOD - On-device sentence embeddings via transformers.js (WebGPU, falling back to WASM). The default model (~30 MB) downloads once from Hugging Face and is then cached for offline use. - In-browser MinHash for self-overlap, PDF text extraction via pdf.js, and IndexedDB for local storage. - Public APIs: PubMed E-utilities, OpenAlex and Crossref. You can add your own free NCBI key for faster PubMed searches. Built for students and researchers who have no budget and no institutional Turnitin/Scopus access — and who deserve good tools anyway.