01 Search plan
Describe your question. Get a search plan you can edit.
Start a search brief in plain words: the topic, the question you are actually asking, what to include and what to leave out. Fihris turns the brief into a set of targeted queries, each aimed at the source most likely to answer it, and gives you the plan to review and edit before it runs. Run it and the results come back merged and de-duplicated across every source at once, ready to add to a collection.
It does not read the papers for you and it is not an autonomous agent: it writes a search plan, you edit it, and you decide what enters your library. Planning always produces something runnable: with no key, no subscription or no network it falls back to on-device query expansion. When a cloud route is used, only the brief is sent, after you agree, and never a PDF.
02 Discover
One search across nine scholarly sources.
Discover queries arXiv, Crossref, OpenAlex, Semantic Scholar, PubMed, Europe PMC, dblp, ERIC and INSPIRE-HEP, merges duplicates, and pulls open-access PDFs straight to the device. Unpaywall looks for legal open-access copies of the results; Springer Nature joins the search only with your own key. Each source keeps its own pace: arXiv gets one request every three seconds on a single connection, which is what its terms ask for.
It does not bypass a paywall. Where there is no open-access copy, Fihris links you to the publisher and stores the metadata.
03 Library
Everything you have read, in one list.
Import a PDF, a BibTeX or RIS file, or anything you find in Safari through Share to Fihris. Papers go into collections you make, tags you invent, and a search that reads the full text of every PDF, not just the titles. Metadata is stored in the CSL-JSON shape the rest of the reference-manager world uses, so moving in and out is simple.
It does not do shared group libraries, and there is no Fihris account to sync them through: your library syncs through your own iCloud or not at all.
04 Reader
A reader that keeps your marks where you made them.
Highlight, underline, strike through and write on the page. Scanned papers get OCR so the text is selectable and searchable. A reflowed reading mode makes a two-column PDF readable on a phone. Every mark keeps the exact span it was made on, so a note always finds its way back to the sentence.
It does not edit the PDF itself. Your markup is kept beside the file and burned in only when you export an annotated copy.
05 Notes
One note per paper, and a Pencil that behaves like one.
Each paper keeps a single note: rich text and Apple Pencil ink on the same canvas, with Scribble, double-tap and squeeze supported where the hardware has them. Select a passage in the paper and it drops into the note with the link back to its page intact.
It is not a second app for general note-taking: notes live with papers, and there is no free-floating notebook.
06 Ask
Answers that can point at a page but cannot invent one.
Fihris splits every PDF itself, so it owns the mapping from a passage to its paper, page and character range. The model is handed numbered passages and may cite only by number; it is never asked for a page number and never permitted to write one. Fihris resolves the numbers back to pages from its own index, and a number nothing matches becomes no citation at all.
It cannot tell you about a paper that is not in your library, and it will say so rather than guess. The choice of passage can still be wrong, which is why the chip opens the page.
07 Mind maps Fihris Pro
Lay a topic out your way.
A mind map is your own structure: topics, branches in colours you choose, and the papers and notes you put under them. Draw it by hand, or have a collection turned into a first outline and take it from there.
It does not derive itself from citation data; that is what Analysis is for, and the two are deliberately kept apart. Nothing in a mind map claims to be a fact about the literature.
08 Analysis Fihris Pro
See how your papers cite each other, and what they all cite that you haven’t read.
Analysis draws who cites whom among the papers you chose, from citation records fetched from the open scholarly sources. Pick any paper and the graph grows both ways (the works it cites, and the works that cite it) and you add what you want as you go. Layouts for time, cluster and hierarchy; and what several of your papers cite but your library is missing shows up as a gap.
It is only as complete as the sources are. A citation no open index records is a citation Fihris cannot draw, and it marks what it could not resolve rather than smoothing it over.
09 Manuscripts Fihris Pro
Write the paper beside the papers.
A LaTeX editor with the library one tap away: insert a citation from the cite picker, render APA, IEEE or Chicago style, and keep the bibliography in step as you write. An approximate live preview shows the shape of the manuscript as you type.
It does not compile a finished PDF on the device: no TeX engine ships with a licence we can use. Export the project, or hand it to Overleaf in one tap, and compile there.
10 Export
Everything leaves in a format something else can read.
Per paper or per collection: BibTeX, RIS, CSL-JSON, a Markdown file of your notes and highlights, an annotated copy of the PDF, and the original. Mind maps leave as PDF, Markdown, OPML, GraphML or DOT. Export everything writes the whole library (papers, PDFs, annotations, notes and drafts) into one archive.
There is no proprietary container to be trapped in, and no export that only Fihris can read back.
11 Every size
The phone is the whole app, laid out smaller.
One universal app for iPhone and iPad, iOS 18 or later. It responds to the size it is given rather than the device it is on, so Split View, Stage Manager and a rotated phone are all just widths. Dynamic Type, VoiceOver and full keyboard control throughout.
Apple Pencil is the one hardware difference. Everything else a large screen does, a small screen does too, in sheets and stacks instead of panes.