Contact person: Business Client

Phone:Show

Email:Show

Location: Singapore, Singapore

Budget: Recommended by industry experts

Time to start: As soon as possible

Project description:
"I have a collection of mixed-content PDFs (rich text with lots of images and the occasional table) that I routinely turn into study notes by following a specific format (Sample given to me by my tuition centre). Doing it by hand is slow, so I’d like a reusable n8n workflow that ingests any of these PDFs and exports a single, fully formatted Word document (.docx) that stays true to the desired layout.

Scope of what the workflow must do
• Watch a folder or accept a manual upload of a PDF.
• Parse text and extract embedded images and tables without quality loss.
• Do the same thing with my format PDF, focusing on preserving structure and formatting.
• Feed both parsed documents into an AI Agent, and output a clean .docx that mirrors the source formatting and includes images, tables, and is ready for immediate study use—no further tweaking on my end.

Technical notes
I am already running an on-prem n8n instance, so please build the flow with standard n8n nodes and, if needed, lightweight Function/Execute Command nodes that call open-source tools such as pdf-miner, tabula, or mammoth for .docx generation. Keep external dependencies well-documented. I am currently using LlamaParse but it has problems preserving images and tables.

Deliverables
1. A ready-to-import n8n workflow JSON.
2. A short README explaining any environment variables, CLI tools, or credentials required.
3. A sample run showing the original PDF and the produced Word file for validation.

Acceptance criteria
• Heading hierarchy, inline styles, images, and tables in the Word output match the source PDF.
• The workflow runs unattended from start to finish with a single trigger.
• All required tools are either bundled or installable with clear instructions." (client-provided description)