Read
Turn PDFs and scanned pages in Indian languages into text you can edit. Today this is THATCHED Convert: PDF in, Word and Excel out, in English, Telugu, Hindi and Tamil.
Live nowTHATCHED Convert: Indian-language PDFs to Word & Excel
AI for every language. Starting with India's.
Turn PDFs and scans in Telugu, Hindi, Tamil and English into editable Word and Excel.
Why this matters
A billion people are still waiting for AI that reads their language.
Forms, land records, invoices, school notices and court papers across India are printed in Indian scripts. The tools that read, search and reuse documents were built for English first. Everyone else retypes.
22
languages listed in the Eighth Schedule of India's Constitution.
Source: Constitution of India, Eighth Schedule.1.4 bn
people live in India, the most populous country in the world.
Source: UN World Population Prospects 2024.~10%
of people in India reported English as a first, second or third language (10.6%).
Source: Census of India 2011, language tables.The platform
Read. Understand. Listen. Build.
One stack of language AI, shipped one honest piece at a time. Each piece carries its real status.
Understand
NextRebuild tables from scanned pages and pull structured fields (names, amounts, dates) out of documents.
Listen & speak
ResearchSpeech recognition and voice for Indian languages, including the way people really mix them with English.
Build
PlannedA developer API so others can put Indian-language reading into their own products.
Live product
THATCHED Convert
Drop a PDF, pick the document language, and get a Word file and an Excel file back. Digital PDFs keep their tables as real Excel cells. Scanned pages are read with OCR in English, Telugu, Hindi or Tamil, alone or mixed with English.
First 2 pages of every file freeNo sign-up. A file of 2 pages or fewer is completely free.
₹100 / extra page₹100 (INR 100) per extra page, paid by UPI in India.
£1.00 / extra page£1.00 (GBP 1.00) per extra page, paid by card outside India.
Deleted on a timerFree files are deleted after 24 hours, paid orders after 7 days.
Never used to train AIYour files are processed, returned and deleted.
Honest limits. Word output is clean, flowing text: images, fonts and multi-column layouts are not kept. On scanned pages we read the text but do not yet rebuild tables into Excel cells. Results are best on clean, printed scans.
Languages
Every script, in its own letters.
Four languages are live in Convert today. The rest of India's 22 scheduled languages are on the roadmap, and we mark them that way.
Live in Convert
PlannedNot available yet
- অসমীয়াAssamesePlanned
- বাংলাBengaliPlanned
- बड़ोBodoPlanned
- डोगरीDogriPlanned
- ગુજરાતીGujaratiPlanned
- ಕನ್ನಡKannadaPlanned
- کٲشُرKashmiriPlanned
- कोंकणीKonkaniPlanned
- मैथिलीMaithiliPlanned
- മലയാളംMalayalamPlanned
- ꯃꯤꯇꯩꯂꯣꯟManipuriPlanned
- मराठीMarathiPlanned
- नेपालीNepaliPlanned
- ଓଡ଼ିଆOdiaPlanned
- ਪੰਜਾਬੀPunjabiPlanned
- संस्कृतम्SanskritPlanned
- ᱥᱟᱱᱛᱟᱲᱤSantaliPlanned
- سنڌيSindhiPlanned
- اردوUrduPlanned
Roadmap
Now, next, later.
No dates we can't keep. This is the order we are building in.
Now
Read
- THATCHED Convert is live: PDF to Word and Excel
- OCR in English, Telugu, Hindi and Tamil
- First 2 pages free, then pay per page by UPI or card
Next
Understand
- Rebuild tables from scanned pages
- Structured extraction from forms and records
- More Indian languages
Later
Speak & build
- Speech in Indian languages
- Developer API
- Languages beyond India
Principles
How we build.
01
Private by default
Files are processed on our own server, never shared and never used to train AI. Each file is deleted on a fixed timer.
02
Priced for India
Pay per page in rupees with UPI. No subscription and no account.
03
Built for phones
Light pages that work on low bandwidth and a phone screen, because that is how most people get online.
04
Honest about limits
We say what works today, what is next and what is still research, and we mark it on every page.
Who we are
A small, technical team.
THATCHED is a UK company building language AI for India. Dr. Tharun Reddy Kandukuri (PhD in Engineering, University of Cambridge) leads product and engineering.
Have a stack of paperwork in Telugu, Hindi or Tamil?
Try the first 2 pages of any file for free. No sign-up.