I built an API to stop manual data entry from invoices and resumes
4 comments
definitely going to pass this on to a couple friends who were just talking about vendor/sales data issues this past week.
Thanks a lot for the support, I'd be happy to support them and offer some free credits to try it.
Why not just use a standard LLM prompt?
You absolutely can for prototypes, but at production scale, you'll hit major issues with cost, latency, and random JSON formatting errors. We handle the heavy lifting—optimizing the vision pipeline and enforcing strict schemas—so you don't have to build and maintain the glue code around the model yourself.
I’m the founder of Scanny AI (https://scanny-ai.com/).
I built this because I noticed that despite all the advancements in AI, businesses are still hiring people to manually copy-paste data from PDFs to Excel. Standard OCR tools often just give you a "blob of text" that still requires manual cleanup.
What it does: Scanny AI takes unstructured documents (Invoices, Resumes, IDs, Receipts) and extracts specific data points into structured formats (JSON, CSV, Excel).
How it works: Unlike regex-based parsers or standard OCR, we use context-aware models to understand the document layout. This means it can identify a "Total Amount" on an invoice even if the layout changes, or extract "Implied Skills" from a CV that aren't explicitly listed as keywords.
Current Use Cases:
Invoices: Extracting line items, tax, and vendor details.
Resumes: Parsing experience and skills for HR.
IDs: extracting PII for KYC checks.
We are currently in Early Access and I’m looking for feedback on the extraction accuracy and the API usability.
I’ve enabled Free Credits for new sign-ups so you can test it on your own documents without paying.
I’d love to hear your thoughts on the edge cases (messy handwriting, weird layouts, etc.) and what features you’d like to see next.
Link: https://scanny-ai.com/
Thanks!