Built by investment professionals, for investment professionals:
Search, screen, analyse documents and create reports in a single place.
Waitlist is open now at one.felixresearch.com, so you can set up access before the app launches.
Data Center revenue reached $115.2B, up +142% YoY1, driven by Hopper GPU demand from hyperscalers and sovereign-AI customers2. Management flagged ~$5.5B at risk from export controls3, partially offset by an H20 re-spec.
It’s the foundation. Every answer, every figure, every summary traces back to a line in a document.
Our ingestion, OCR, and entity layer are our own. We control the primitives so the quality compounds.
FelixOne behaves like a seasoned analyst: precise, skeptical, and allergic to filler.
Felix One groups every document, search and report under a single workflow. Four steps from a blank workspace to a memo your IC will read.
Everything in Felix One lives inside a workflow: a deal, a thesis, a quarterly LP letter. Create one and you get a shared library, a knowledge graph stitched across the documents, all of your related chats and every report you have drafted from it. Each of the four steps below runs inside one.
Drop your CIMs, filings, lender decks, transcripts and management presentations. Felix OCR runs on PDFs and images, then a proprietary extraction pass surfaces entities, tables and the relationships that tie them across the workflow. Pre-defined outputs: diligence briefs, target metrics, comparable tables, start running the moment a file lands.
Three ways to interrogate the same corpus. Run a grep-style search across the workflow when you know the phrase. Pull a pre-defined summary or metric when you want the headline. Talk to Felix Intelligence when the question is open: Felix walks the knowledge graph across documents and returns an answer with every claim anchored to its source.
Turn your reading into output. Felix drafts IC memos, diligence briefs, valuation notes and LP letters from the workflow library; every paragraph carries a citation back to the document that supports it. Documents are editable in-app: rewrite a paragraph and the citations stay attached, drop in your own commentary, share or export when it is done.
Atlas is a mid-market industrial distributor generating $420M of revenue at a 14% EBITDA margin1. The sponsor proposes a multi-site roll-up, with three bolt-on candidates identified in the management presentation2.
Net leverage stands at 3.8× at FY25 close, with the maintenance covenant set at 4.25× through FY273. A springing covenant on net leverage triggers a cure right above 4.5×4.
Export to PDF, Word or Excel with citations and underlying tables intact, or share a read-only link with your team. Felix routes the memo onward: into the IC distribution by email, as the matching LP letter draft, or as the order ticket the trading desk needs. Pick the next step, Felix carries the work over.
Felix One is moving quickly. These surfaces are in active build; early access opens to existing workflows first.
Visit the Felix One app site, create your account, and we will open access from there as Build 1.0 rolls out. The product is not fully live yet, but account creation is open now.
We won’t pretend nothing leaves your laptop. OCR and entity extraction need real compute, so pages do travel to our servers — briefly, under contract, and never to train a model. The parsed index is stored on your device. Your queries against it run there too.
Pages are sent to our infrastructure for OCR and entity extraction. The temporary upload is deleted as soon as the result reaches your browser — and in any case within 24 hours, whichever comes first.
Once a document is parsed, the structured result — the index you actually search and cite against — is stored securely on your device. Search runs locally; we don't see your queries against your own corpus.
When a question goes to a model, it travels through providers bound by zero-retention, no-training agreements — and we're actively moving toward sealed routes (e.g. Amazon Bedrock). Nothing you upload or ask is ever used to train a model.
For anything not covered here, email hello@felixresearch.com. James reads everything.
Three features, shipped and stable: Intelligent Search, Document Analysis (with the citation layer), and the security/compliance baseline. Collaboration, integrations (Claude/Codex/API/CLI), and live market data are on the bench and shipping through H2–Q4 2026.
PDFs (text or scanned), plain text, HTML, and links — paste a single SEC URL or an entire investor-relations site and we'll fetch every filing on it and run the lot through the pipeline. Scanned PDFs are OCR'd automatically. No hard page cap on the institutional plan.
About two minutes end-to-end on a warm pipeline for a typical 200-page 10-K. Very dense documents with many tables can take a bit longer. You can start asking questions the moment structure-parsing completes — entity linking finishes in the background. Faster processing options will be available for enterprise customers.
We built the extraction layer in-house, specifically for financial filings. Tables are parsed as tables, footnotes stay attached, multi-column layouts are reconstructed, and comma-formatted numbers are normalised. If you hit a document where it struggles, send us the file — we'll fix it.
Pages travel to our EU-hosted infrastructure for OCR and entity extraction. The temporary upload is auto-deleted as soon as the structured result reaches your browser, and in any case within 24 hours. The result itself — the index you search and cite against — is stored securely on your device. Your data is never used to train a model, ours or a provider's.
H2 2026, alongside the Claude/Codex tools and the CLI. The four “Anywhere” surfaces ship as a single quarter — they share the same core. Early-access list is open; tell us which surface matters most and we'll prioritise your slot.
Felix One is tiered by seat with volume discounts for teams of 10+. The full schedule lands when general availability opens — in the meantime, get in touch and we'll talk through what fits.
Subscribe for Felix Research notes, product updates, and launch writing from the team building Felix One.
Subscribe to The Felix View for research notes and launch updates.
