ExtractFox
Author profile

Dawid Sibinski

Founder and engineer at ExtractFox. Writes about AI document extraction, multimodal models, OCR vs IDP, and structured-output patterns.

About

I build ExtractFox, an AI-native data extraction tool that turns PDFs, images, and scanned documents into structured data — without per-vendor templates. Most of what I write is engineering notes from doing that, plus workflow guides for the people who use it.

For background reading, the glossary covers the terms (OCR, IDP, multimodal AI, MRZ, structured output) that keep coming up in the posts below. More about ExtractFox.

Posts (57)

All posts →
Tutorial5 min read
How to extract data from a chart in Excel
Get the underlying numbers back out of an Excel chart — even if the source range was deleted, the workbook is locked, or you only have the chart as an image.
Engineering5 min read
How to extract metadata from a PDF
Author, creation date, producer software, custom XMP fields — every PDF carries metadata most people never see. Here's how to read it, in any language, and what's worth pulling for indexing or audit.
Tutorial5 min read
How to extract images from a PDF
From a one-off Preview save to batch extraction with pdfimages, pypdf, or iText — every realistic way to pull every embedded image out of a PDF at full quality.
Engineering5 min read
How to extract links from HTML or plain text
BeautifulSoup, Cheerio, PHP DOMDocument, and the regex you should and shouldn't use — every reliable way to pull URLs out of a string of HTML or text.
Tutorial4 min read
How to extract a zip file (Mac, Windows, Linux, Android)
What "extracting a file" actually means, the one-click way to do it on every major OS, and how to handle the file that won't open.
Engineering6 min read
How to extract keywords from text, a website, or a job description
RAKE, YAKE, KeyBERT, TextRank, and the LLM era — the four open-source keyword-extraction algorithms that still matter, plus when to use each.
Engineering5 min read
How to extract sentiment from text
VADER, TextBlob, fine-tuned transformers, and the LLM-with-typed-schema pattern — four sentiment-extraction methods, what each one is best at, and the few lines of code to start.
Workflow5 min read
How to extract a chart of accounts from QuickBooks
Three reliable ways to get your full chart of accounts out of QuickBooks Online or Desktop — built-in export, IIF, and the API route — plus how to clean it up before re-importing somewhere else.
Workflow6 min read
How to extract data from a website to Excel automatically
Power Query, Make/Zapier, and a no-code AI route — three ways to set up an automated pipeline from a website to an Excel file that updates on its own.
Tutorial5 min read
How to extract data from a pivot table in Excel
GETPIVOTDATA, Show Details, copying values, and converting a pivot back into a flat table — the four ways to get data out of an Excel pivot, and when each one is the right call.
Tutorial4 min read
How to extract RAR, tar.gz, jar, and other archive formats
Beyond zip — every other archive format you'll meet (RAR, 7z, tar.gz, gz, jar, war, ear) and the right tool for each on every OS.
Engineering5 min read
How to extract topics from text or interview transcripts
BERTopic, LDA, and the LLM-with-clustering pattern that's quietly taken over qualitative research. What each one is best at, and the few lines of code to start.
Tutorial6 min read
How to extract data from Zillow listings
Zillow doesn't offer a public scraping API and actively blocks bots. Here's how to get listing data into a spreadsheet without getting your IP banned — and the legal lines to stay behind.
Engineering6 min read
How to extract all links from a website
Crawling a whole site for every internal and external URL — the right way with a sitemap, the brute-force way with Scrapy, and what to do when neither works.
Tutorial5 min read
How to extract photos and frames from a video
Pull a still from a video — on iPhone, Android, in DaVinci Resolve, with FFmpeg, or from a YouTube URL. The right tool depends on whether you want one frame or every frame.
Tutorial4 min read
How to extract text from a PowerPoint file
Three ways to pull text out of .pptx files — the built-in outline view, scripting with python-pptx, and image-based extraction for slides where the text is baked into pictures.
Tutorial5 min read
How to extract EXIF and GPS metadata from a photo
Where the photo was taken, what camera made it, when it was shot, even the focal length — every JPEG carries it. Here's how to read EXIF from a photo and the privacy lines you should know about.
Engineering7 min read
How to extract a table from a PDF using Python
pdfplumber, Camelot, Tabula, and the API-based fallback — what each library handles well, what it breaks on, and the code you actually need.
Tutorial5 min read
How to extract images from a website (or a URL)
Browser tools, wget, gallery-dl, and the legal lines around scraping images from sites like Instagram, Pinterest, and stock photography. What's safe, what's gray, and what to skip.
Tutorial4 min read
How to extract metadata from a video file
FFprobe, MediaInfo, and yt-dlp — three tools that cover every format from MP4 to MKV to a YouTube URL. What each one is best at, and what you can pull out.
Engineering5 min read
The end of templates: how AI extraction actually works
Per-supplier templates were the only way to extract structured data from PDFs for two decades. Multimodal models change the shape of the problem.
Tutorial4 min read
How to extract hyperlinks from Excel and Google Sheets
Excel hides hyperlinks behind display text — getting the actual URL out takes a HYPERLINK trick or a tiny VBA function. Google Sheets has its own quirks. Here's the full set.
Tutorial4 min read
How to extract images from Google Docs and Google Slides
Three reliable ways to get every image out of a Google Doc or Slides deck — with or without losing resolution — including the publish-to-web trick that beats every other method.
Tutorial5 min read
How to extract a signature from a PDF or an image
Whether you need to verify a signature exists, lift it as a transparent PNG for reuse, or pull every signed name as text — here are the right tools for each version of the question.
Tutorial3 min read
How to extract images from a Word document
The fastest way to pull every image out of a .docx file at original resolution — using nothing more than the file extension trick that works in any unzip tool.
Tutorial7 min read
How to extract numbers from a cell in Excel
Whether you need digits out of a product code, an order ID, or a free-text field, here are the formulas (old and new), the Power Query route, and what to do when the data isn't actually in Excel yet.
Tutorial5 min read
How to extract metadata from a website
Title tags, Open Graph, Twitter cards, JSON-LD structured data — what every page exposes and how to pull it out cleanly for SEO audits, link previews, or content indexing.
Tutorial4 min read
Built-in image-to-text features in Mac, OneNote, and Excel
Live Text on macOS, OneNote's Copy Text from Picture, and Excel's Data from Picture — the OCR features already on your machine that most people don't know exist.
Tutorial4 min read
How to extract a table from a Word document
Native Word-to-Excel paste, python-docx for scripting, and what to do when the table is actually a screenshot inside the document.
Tutorial4 min read
How to extract embedded files and attachments from a PDF
PDFs can carry attached files — Excel sheets, source data, supporting docs. Acrobat shows them; most other readers don't. Here's how to get them out, on any OS.
Tutorial4 min read
How to extract zip and postal codes from addresses (Excel, Sheets, Python)
Three reliable patterns for pulling postal codes out of free-text addresses — Excel formulas for clean US data, regex for international, libpostal when nothing else works.
Tutorial4 min read
How to extract video links from a YouTube playlist
Three ways to get every video URL from a YouTube playlist into a flat list — yt-dlp for scripts, the official API for production, and the browser console for one-offs.
Tutorial5 min read
How to extract phone numbers from text
A regex that handles most cases, the libphonenumber library that handles the rest, and what to do when the phone numbers are trapped inside PDFs, screenshots, or messaging exports.
Engineering6 min read
How to extract key-value pairs from documents
"Name: John Smith. Date: 2024-04-12. Total: $1,420." — every form, invoice, and structured PDF is full of key-value pairs. Here's how to pull them out reliably across formats.
Engineering4 min read
How to extract files from a Docker image or container
docker cp from a running container, dive for digging into image layers, and the few tricks for getting files out of an image you can't run.
Engineering4 min read
How to extract the host or domain from a URL
URL parsing in Python, JavaScript, Bash, and SQL — and the public-suffix-list trick that makes "co.uk" come out right.
Tutorial5 min read
How to extract text from a YouTube video
Three reliable ways to turn a YouTube video into searchable, citable text — using the built-in transcript, yt-dlp + Whisper, or browser tools — and when each one is the right call.
Engineering5 min read
How to extract city, state, and country from a location string
"San Francisco Bay Area," "Greater London," "NYC." Free-text location fields are messy. Here's how to parse them into clean city/state/country with libpostal, geopy, or an LLM.
Engineering5 min read
How to extract schema from SQL Server, MongoDB, JSON, XML, or Parquet
Information_schema, sp_help, mongoexport, jsonschema, parquet-tools — the right command for every common data store, in two lines or fewer per format.
Tutorial4 min read
How to extract formulas from an Excel file or PDF
Showing all formulas in a sheet, exporting them programmatically with openpyxl, and pulling math from a PDF where the formulas are rendered images, not LaTeX.
Workflow5 min read
How to extract key information from emails
Sender, dates, attachments, action items, deal numbers — emails are unstructured text wrapping structured data. Here's how to pull the structure out reliably for CRM, support, or finance workflows.
Workflow4 min read
How to extract a Gantt chart from MS Project
Get a Project file's tasks and timeline out into Excel, image, or a presentation-friendly format — without buying a copy of MS Project if you don't already have one.
Tutorial4 min read
How to extract code from a video tutorial
Three workflows for getting the source code out of a programming tutorial video — from manual frame capture to a full transcript-plus-screenshot pipeline.
Workflow5 min read
How to extract an organization chart from Microsoft Teams
Teams shows you reporting lines but doesn't give you a clean export. Here's how to get the org structure out via Microsoft Graph, the People app, or as a fallback, from screenshots.
Tutorial5 min read
How to extract a summary from a PDF, video, or article
Three classes of summarization tools (extractive, abstractive, hybrid), how each one fails, and the practical setup for getting a useful 10-line summary out of a 90-minute talk or 300-page report.
Tutorial6 min read
How to extract data from Amazon product pages
Title, price, ASIN, ratings, variations — Amazon makes it hard to scrape and easy to misuse APIs. Here are the legitimate options, the gray ones, and the screenshot-based fallback.
Workflow3 min read
Slack's auto-extract links setting (and what it actually does)
Slack quietly fetches link previews for every URL pasted in a channel. Here's how the setting works, why teams turn it off, and how to control it per-channel or per-message.
Workflow5 min read
How to extract a chart of accounts from SAP
SAP exposes the chart of accounts through a few transactions and a few APIs — here's the practical map for finance teams who need a clean export, not a consultant's project.
Engineering5 min read
How to extract facts from text
Named entities, claims, relations — what "fact extraction" actually means in NLP, the libraries that handle each piece, and how the LLM era changed which ones are worth using.
Workflow4 min read
Closing the month without re-keying invoices
A practical workflow for accountants and bookkeepers: take a folder of supplier PDFs, get a clean Excel sheet, and reconcile in one sitting.
Engineering7 min read
How to extract LinkedIn profile data using Python
LinkedIn is the most legally fraught scraping target on the internet. Here's what's allowed, what's gray, and the safest patterns for engineers who need profile or company data programmatically.
Engineering7 min read
How to extract text from an image using Python
Tesseract via pytesseract, EasyOCR, PaddleOCR, and the API-based path — what each one is best at, what they break on, and the few lines of code to get started.
Industry6 min read
What "unstructured data" actually means — and how to extract from it
Unstructured data isn't disorganized data — it's data without a schema. Here's the practical taxonomy, why it became extractable in the last few years, and the patterns that work.
Workflow4 min read
How to extract questions and responses from a Google Form
Three ways to get Google Forms data out — the built-in Sheets export, the Forms API for automation, and the screenshot fallback for when you only have access to the published form.
Engineering7 min read
How to extract data from a PDF in C#
A working engineer's tour of the PDF extraction libraries in the .NET ecosystem — iText, PdfPig, Azure Document Intelligence, and the API-first alternative when you don't want to ship a parser at all.
Workflow5 min read
How to extract data and metadata from PDFs with Power Automate
The built-in actions for PDF text extraction, the AI Builder model for invoices and receipts, and how to wire either one into a flow that drops structured data into Excel or Dataverse.
Product3 min read
Why we don't store your documents
A short note on how ExtractFox processes files in-flight, what we keep, and what we throw away — and why that's a deliberate product decision.