Upload PDFs, Word, PowerPoint, Excel, CSV, images or a website URL — then ask anything. Exact numbers from your spreadsheets, cited paragraphs from your documents, structured fields from your invoices.
Exact answer · SQL on 2,418 rows
| North | $412,380 |
| West | $365,910 |
| South | $298,045 |
| East | $251,700 |
“Your team works across PDFs, Word, slides, Excel, CSVs, scanned invoices and websites — and every format needs a different tool, a different person, or a VLOOKUP that breaks at midnight. Important answers live in the files you already have. You just need to ask them.”
Files2BI answers them — with the source attached.
PDF, Word, PowerPoint, Excel/CSV, images, or paste a URL. Files2BI reads each one and works out its structure.
For spreadsheets and CSVs it builds a real column schema (types, samples, distributions) so questions are answered with SQL, not guesses.
For PDFs and Office files it extracts text, tables and form fields (invoice number, total, due date) so both prose and field-level questions work.
Every answer is cited — page number, row count, matching documents, or invoice number with a one-click Open button.
Open Visual Dashboards on any tabular file for auto-suggested charts, or build your own from a prompt or by picking columns.
One tool for every business file — with answers you can check.
Before: Tabs across Excel for spreadsheets, a PDF reader for documents, an OCR tool for invoices, and a search bar that doesn't work.
With Files2BI: Accepts PDF, Word (.docx/.doc), PowerPoint (.pptx/.ppt), Excel (.xlsx/.xls), CSV, text, Markdown, images (PNG/JPG/WEBP/GIF) and website URLs — all queryable from one chat box.
Before: VLOOKUP, SUMIF, COUNTIF, INDEX/MATCH and pivot tables break when columns move or someone deletes a row.
With Files2BI: Ask “sum revenue by region where status is paid” or “average downtime per machine” — it runs as real SQL against your rows. No #REF!, no rebuild when the data updates.
Before: AI chat tools guess at numbers when asked to total a 50,000-row CSV, and bury the wrong answer in confident prose.
With Files2BI: XLSX and CSV rows are loaded into PostgreSQL with type-classified columns. Sums, averages, medians, percentiles and standard deviations run as SQL — exact numbers, every time.
Before: Building a chart in Excel takes twenty minutes and nobody shares it.
With Files2BI: Open any tabular file to see auto-generated charts (time series, top-N breakdowns, KPI tiles). Build custom charts from a prompt or by picking columns: bar, line, pie, KPI, scatter, table.
Before: Field-level questions about invoices, contracts and forms force someone to open every PDF.
With Files2BI: Reads invoice number, date, total, vendor, line items, due date, tax ID and more automatically. “Show all Salesforce invoices over $5,000” returns invoice cards with Open buttons.
Before: Answers often span many files — opening 30 PDFs to compare clauses is half a day of work.
With Files2BI: Ask across all your documents at once. Only the relevant passages of each file are retrieved, and the answer cites each document it used.
Card numbers, bank accounts, government IDs, contact details and medical record numbers are detected and masked before anything is stored, indexed or sent to the AI — so they never appear in answers, tables or exports. Card and account numbers keep their last four digits so you can still tell them apart.
Masking is automatic and best effort: it can miss values in unusual formats, images or handwriting, so treat it as a safeguard, not a guarantee. See section 7 of our Terms of Service.
Masked automatically
Uploaded: card 4111 1111 1111 1111, a/c 50100234567890
Stored: card [CARD ••••1111], a/c [ACCOUNT ••••7890]
Person and company names, amounts and dates stay readable.
Finding one clause across hundreds of contracts takes hours of manual searching.
Ask “what are the termination clauses across all vendor contracts?” and get cited answers in seconds.
New hires ask the same policy questions again and again — each answer takes 20 minutes.
Upload the handbook once; everyone asks it themselves and gets page-level citations.
Audit prep means reading regulatory documents end to end to find the right sections.
Ask across all compliance documents at once; every answer cites the source page.
SOPs, downtime logs and shift reports live in separate spreadsheets, PDFs and folders.
One place to ask — “median downtime per shift in week 14” works on the Excel, “what does the SOP say about lockout?” works on the PDF.
Sensitive contracts? Browser-Private mode keeps the files on your computer and the search index in your browser — nothing is stored on our servers.
Legal & Compliance
Hundreds of contracts per client, each 50+ pages. Finding specific clauses during due diligence or litigation takes hours.
“What are the termination clauses across all our vendor contracts?”
Healthcare
Clinical protocols, insurance guidelines and compliance documents scattered across folders.
“What is the protocol for administering medication X to patients with condition Y?”
Education
Curriculum documents, accreditation handbooks and policy manuals hundreds of pages thick.
“What is the procedure for handling student appeals in our grading policy?”
Manufacturing
Technical manuals, equipment documentation and safety SOPs thousands of pages thick.
“What are the calibration steps for the CNC machine model X?”
Real Estate
Lease agreements, property disclosures and HOA documents that have to be re-read for every question.
“What are the pet policies across all rental properties we manage?”
Start with 50K free tokens. Then buy token packs from $5 — no subscription, no seat fees, tokens never expire.
Starter
$5
400K tokens
Professional
$50
5M tokens
Enterprise
$250
35M tokens
PDF, Word (.docx/.doc), PowerPoint (.pptx/.ppt), Excel (.xlsx/.xls), CSV, plain text, Markdown and images (PNG, JPG, WEBP, GIF). You can also paste a website URL and Files2BI reads up to 10 pages on that site.
Spreadsheet and CSV rows are loaded into a SQL database with typed columns. When you ask for a total, an average or a percentile, Files2BI runs a SQL query on your rows — the AI writes the query, it doesn't estimate the number.
Every answer cites its source: the page of the document, the rows and filters of the spreadsheet, or the invoice it came from, with a button to open the file.
In Cloud mode, uploads are stored encrypted in Amazon Web Services (US East) and are only accessible to your workspace. In Browser-Private mode the files stay on your computer and the search index in your browser: their text is sent to us only to be masked and indexed with Amazon Bedrock, and nothing is stored on our servers.
They are masked automatically when the file is uploaded: card and bank account numbers, IBAN, IFSC and routing numbers, CVV and expiry dates, SSN, Aadhaar, PAN and passport numbers, emails, phone numbers, postal addresses, dates of birth and medical record numbers. The masked values are never stored in your workspace, sent to the AI or shown in answers; card and account numbers keep their last four digits so you can still tell them apart. Person and company names are not masked. Your original file stays private to your workspace and downloads unchanged. Masking is automatic and best effort — it can miss values in unusual formats, images or handwriting — so it is a safeguard rather than a guarantee (see section 7 of our Terms of Service).
More answers on the FAQ page.
Start with 50K free tokens. No credit card. No commitment.
