Woooow Mark, this is amazing!
So you built this prototype for all your personal data not only for cars?
I have been thinking about using OCR AI packages to automatically read data from the upload, either a photo or a pdf, and only allow minor changes from the car owner to correct mis-read.
How are these data input now? manually typing?
I built some frameworks for Cars, Trips, Medical data, Pensions, etc.
The value seems to be in:
a) a repository for one or two random bits of data like NI number and Premium Bond 'holder number' (of course this is UK-centric)
b) a repository for ongoing lists under a heading, like car history
What I have done is all manual entry (in a spreadsheet), and I have not really kept it up. I am nerdy enough to want to keep all this data somewhere, but too lazy to actually do it.
Scanning is easy, especially if you have a good scanner, like the Fujitsu Scansnap that I use.
OCR is fairly simple, although some formats (notably Barclays Bank statements) are very troublesome.
But 'understanding' the OCR is much harder. But at 'level 1', if I were to tell the system what it is getting (Expedia Itinerary, MOT Certificate, or whatever, then it would be pretty simple to allocate the OCR data to fields, I suppose. Someone would have to build loads of import filters. This certainly exists (and works pretty well) for scanning receipts into expenses systems, but would need to be much broader. And international, and multilingual.
The big issues are:
- Security
- Import Filters
- Data Model
- Market Momentum
I am a little surprised that nobody seems to have put something together that's is interesting enough to raise $100,000 and then $5m and then sell to Microsoft .