Mailhook and Knowledge: Property Disclosure PDF Extraction Template
A disclosure PDF emailed to a Spojit Mailhook address is read, embedded, and queried by AI to pull defects, dates, and seller statements into clean structured fields.
What It Builds
This template starts with a Mailhook trigger, so the workflow gets its own dedicated email address. When a disclosure PDF lands there, an Attachment node fetches the file bytes and a Knowledge node embeds it into a transient collection for that single run. A second Knowledge node in Query mode then reads the document and returns the defects, key dates, and seller statements as structured fields you can route onward.
The Prompt
Paste this into Miraxa and it builds the workflow, connecting the tools for you:
Build a workflow with a Mailhook trigger so people can email a property disclosure PDF to a dedicated address. Fetch the attached PDF, embed it into a transient Knowledge collection for this run, then query it with AI to extract the disclosed defects, important dates, and seller statements into structured fields: property_address, disclosure_date, listed_defects, repair_history, and seller_notes.Connectors Used
- Mailhook - the trigger; gives the workflow its own email address that starts a run on every inbound message.
- Attachment - fetches the disclosure PDF bytes from the inbound mail.
- Knowledge - embeds the PDF into a transient collection, then queries it in Query mode to extract the fields.
Customize It
Change the field names in the prompt to match your own intake form, add fields like hoa_status or environmental_hazards, or attach a Response Schema so the output is always strict JSON. You can also add a Send Email step to forward the parsed summary to your transaction coordinator.
Tips
- The Attachment node only works in Mailhook workflows; use Single mode when one PDF arrives, Multiple when several are attached.
- A transient Knowledge collection embeds, queries, and discards within the run, so nothing lingers between submissions.
- Knowledge handles OCR, so scanned or photographed disclosures still extract cleanly.