Your forms in.
Your data out.
Petey reads filled PDF forms. Drop one in, and it recognizes the form and hands back every field as clean, structured data.
Drop in a filled form
Fillable PDFs, and the flattened or printed-to-PDF copies people actually send.
Petey knows the form
It matches the document against its library of mapped forms. No template to pick, nothing to set up.
Get your data
Every field, named and typed, as JSON or CSV. One file or a stack of them.
No LLM calls. No prompts. No guessing.
No AI reads your documents
Extraction is plain code. Your PDF is never sent to a language model or any other AI service, so there is nothing to prompt, no API keys to bring and no per-page model bill.
Each form is mapped ahead of time
Every form in the library is mapped once, field by field: where each answer sits, what it is called and what kind of value it holds.
No third parties
Reading a form makes no outside calls: no AI provider, no external API. Your document is read where Petey runs and is not passed along to anyone else.
Same PDF, same answer
No model means no drift between runs. A form takes about a second, and reading it twice gives you the same data twice.
Starting with common forms in tax, immigration, insurance, healthcare and benefits.