Hire an AI scanned document reader and fact checker

Everything it can hand back

Your task sets the actual outputs. The specialist picks the methods that cover them.

• The extracted text, with page numbers
• A fact table: each fact, its value, its page and whether it was confirmed on the page
• A list of uncertain readings, with the page and what the text might say
• Checks on the table for missing rows, mixed formats and sums
• A hand-back summary and one next step

How it does the job well

• Turn every request into owned actions: who, what and by when.
• Work from the real calendar, notes and lists. Never invent meetings, people or dates.
• Put the most important decision or deadline first, then the details.
• Prepare drafts and proposals. Sending, booking and changing records waits for approval.

Mistakes it avoids

• Losing commitments buried in long notes.
• Mixing confirmed facts with guesses about other people's plans.
• Adding process when a short checklist would do.

What a good result looks like

An example, not your task: You have a scanned contract, invoice, form or letter and need certain facts out of it.

• Extracted text. The text of the scan, kept in page order with page numbers.
• Fact table. Each fact you asked for, its value, its page and whether it was confirmed on the page.
• Uncertain readings. Places where the text could not be read with confidence, with the page and likely options.
• Table checks. Missing rows, mixed formats and sums that do not add up.
• Hand-back summary. Files saved, MISSING lines, checks that could not run and one next step.

How the work is checked

• Every name, date and number in the fact table was checked against the page image, or is marked uncertain.
• Every fact has a page number.
• No digit was guessed or corrected to make a total work.
• The original scan is untouched, and every file made has a new name.
• States which checks passed, failed or could not run, and names missing inputs.

These checks have not been run against a model yet. Your AI uses them to check its own work.

Capabilities

• Scanned PDF text reader. Reads the scan with OCR. Keeps page numbers and marks readings it doubts. Hands the text and the doubtful readings to step 2.
• PDF maker and reader. Renders the original pages and checks each important fact on the page image, not only in the extracted text. Saves the fact table as a CSV or workbook and hands confirmed facts and open readings to step 3.
• Spreadsheet checker. Takes the saved fact table and checks it for missing records, mixed formats and totals that do not add up.

How the role works

1. Checks your task, context and available tools.
2. Picks the methods needed and prepares a work plan.
3. Carries out supported tasks, with each method's limits and review gates.
4. Checks the real deliverables and corrects supported faults.
5. Hands back files, evidence and unresolved gaps.

Your specialist's working setup

Download the base specialist without a task form or account. README.md and SETUP.md explain how to use the folder. SOUL.md defines its scope. INTAKE.md asks missing questions inside your AI. PLAYBOOK.md, SKILLS.md, BEST_PRACTICES.md, CHECKS.md and HANDOFF.md guide the work and its checked deliverables.

Complete associated skills, supporting files and licences are included. Choose only the methods needed for the actual task. Keep context, working files and accepted feedback in your AI project.

Use this specialist through MCP or the API

Call prepare_specialist with role_id extract-a-scanned-document-and-verify-key-facts only. No task is required to load it. Your AI asks what job you need done, what you want back and which relevant context it can use.

For task-led hiring, call find_specialist, explain the suggested specialists and let the user accept or change the team.

Complete ZIP: Download base specialist. API: GET /api/v1/specialists/extract-a-scanned-document-and-verify-key-facts/pack. The host AI supplies its own model, tools and execution; a pack does not connect accounts or run a task.

Inputs and helpful context

The specialist starts from the task brief. Only active methods require their inputs.

For the primary method, when used

• The scanned file (PDF or image)
• Permission to process it, and whether it must stay on your computer
• The exact facts or fields you need, such as names, dates, amounts and reference numbers
• The pages that matter (optional)
• The language of the document, if it is not English (optional)

Helpful context (optional)

• A second copy of the document, such as a typed version, to compare against.
• A named person who confirms the uncertain readings.

Tools and access

• OCR software on your computer: Scans need an OCR tool and the right language data. Keeping it on your computer keeps private documents private.
• A PDF page renderer: Page images let the AI check each fact by sight. Text extraction alone cannot show layout or stamps.

Functions and deliverables

• Scanned document reader and fact checker: Read text from a scan. Check key names and numbers against the original pages and flag what is unclear.
  • The extracted text, with page numbers
  • A fact table: each fact, its value, its page and whether it was confirmed on the page
  • A list of uncertain readings, with the page and what the text might say
  • Checks on the table for missing rows, mixed formats and sums
  • A hand-back summary and one next step
• Scanned PDF text reader: Give it PDFs, Office files, or scans. Get page-positioned text and OCR using local tools, with optional OCR services.
  • Text from every page you chose
  • The box position of each piece of text, in JSON
  • OCR text for scanned pages
  • Page images for charts or handwriting (optional)
  • A list of any pages that failed
  Source: K-Dense.
• PDF maker and reader: Read a PDF or make a new one, then check every page as an image. Catch cut-off text and layout slips in your PDF before you send it.
  • The text pulled out of your PDF, as plain text, one block per page.
  • A PNG image of each page, named by page number, so you can see the layout.
  • A new PDF saved in output/pdf/ with a stable file name, if you asked for one.
  • A layout check after each change, with any problems named.
  Source: OpenAI.
• Spreadsheet checker: Check a spreadsheet for odd totals, broken formulas and messy data. Your original stays safe, and you get findings with proof and a new file.
  • A sheet inventory and scope
  • Findings with evidence locations
  • Data quality and formula issues
  • A new output file with your edits
  • Reconciliation results
  Source: Burhan Sebin.
• me.md builder: Answer a few questions, or let your AI use what it remembers. Get one private file that tells any AI who you are.
  • One me.md file with up to ten dated sections
  • A list of facts taken from memory, for you to check
  • A short note of what it removed for privacy
  • The two next steps, to save it and set it up

Tasks you can use it for

• You have a scanned contract, invoice, form or letter and need certain facts out of it.
• You need names, dates and amounts in a table, with the page each one came from.
• An earlier text copy of the scan has typos and you do not know where.
• Someone will rely on these facts, such as an auditor, a lawyer or a customer.

Limits to check before starting

• The PDF is already text you can select and copy. Use the PDF maker and reader.
• You want a summary, not exact facts. For a contract, use the Contract plain-language summary writer prompt.
• The document is handwritten. OCR is weak on handwriting. Use this for a first reading and check each fact yourself.
• The document must not be uploaded anywhere. Say so at the start so the AI keeps the work on your computer, or do the reading yourself.

Best way to use this role

Give it one clear outcome, the actual files and a definition of done. Keep the work in one AI project so it can reuse checked context and earlier files.

• Download the handoff pack. It is a zip with HANDOFF.md and every skill file this workflow uses.
• Give the folder to Claude, ChatGPT or Gemini and say "Follow HANDOFF.md".
• Attach the scan. List the facts you need and say whether the file must stay on your computer.
• Read the uncertain readings first. Open the original page and settle each one yourself.
• Check every amount and date in the fact table against the page before you use it.
• Without OCR tools, your AI reads the page images directly and marks each reading to check.

Read it as data

Read the complete role setup as JSON. Connect your AI to the AI Blocks MCP library and read aiblocks://roles/extract-a-scanned-document-and-verify-key-facts for its role instructions and pinned sources.

This is a prepared AI role. No task has run. Your AI carries out authorized work using its available tools and reports the actual deliverables and checks.