Local processing
Keep sensitive PDFs on infrastructure you control.
RelayWorks AI Document Processing Kit
Start Building Instead of Integrating.
A self-hosted FastAPI + Next.js application that turns PDFs into AI-ready output packages while keeping processing on your own infrastructure.
Developers waste days assembling document ingestion pipelines.
PDF upload handling, subprocess execution, background state, output review, error handling, and ZIP packaging all have to work together before the AI work can begin.
Why RelayWorks
A practical application foundation for local document processing workflows.
Complete workflow
Upload, process, inspect outputs, and download packages in one app.
Implementation foundation
Start from FastAPI, Next.js, background jobs, and bounded local uploads.
Source code ownership
Modify the implementation instead of relying on a black-box hosted tool.
How it works
Three steps from local PDF upload to downloadable AI-ready package.
01Upload PDFs
02Process
03Download complete package
Outputs
Generate files that are easier to inspect, package, and use in AI workflows.
Markdown
HTML
JSON
Text
Images
Processing Report
ZIP Package
Privacy
Processing happens locally.
RelayWorks is designed for local or self-hosted operation. Uploads, generated outputs, job metadata, and downloadable packages stay in the local project storage paths you configure.
Who it is for
Technical buyers who need document processing infrastructure for AI workflows.
AI engineers
Technical founders
Internal tools teams
Agencies
Document-heavy RAG teams
Requirements and known limitations
Clear setup expectations before purchase.
Python
Required for the FastAPI backend.
Node.js
Required for the Next.js frontend.
Marker installed separately
The backend invokes marker_single as an external CLI.
PDF-focused
This release supports the existing PDF workflow.
Local setup required
This is not a hosted SaaS.
Screenshots
Real screens from the current product.
Demo
Watch the local PDF-to-package workflow.
See RelayWorks process a PDF locally and generate structured, AI-ready output files.
FAQ
Practical answers for developers evaluating the kit.
What does RelayWorks process?
The current product workflow is PDF-focused.
Is RelayWorks a hosted SaaS?
No. RelayWorks is a self-hosted source-code product.
Does it include source code?
Yes. The FastAPI backend and Next.js frontend source are included.
Does it include Marker?
No. Marker must be installed separately and available through the configured marker_single command.
Does it include model weights?
No. The kit does not distribute Marker model weights.
What outputs are generated?
Markdown, HTML, structured JSON, plain text, extracted images when available, processing reports, and ZIP packages.
Can it process batches?
Yes. The current app includes a batch PDF workflow.
Does it run background jobs?
Yes. Single PDF conversion uses background job processing.
Does it do OCR?
Do not assume OCR beyond what the separately installed Marker workflow supports in your environment.
Can I deploy it locally?
Yes. The product is designed for local or self-hosted operation.
Is an API key required?
No API key is required for the current local workflow.
Who is it for?
AI engineers, technical founders, internal tools teams, and agencies building document-heavy workflows.
What are the main requirements?
Python, Node.js, npm, and Marker installed separately.
Can I customize it?
Yes. The product includes source code intended for buyer modification within the purchased license terms.
Will every PDF convert perfectly?
No. Extraction quality depends on the source PDF, document quality, Marker behavior, and local setup.
Pricing
One source-code kit for local PDF processing workflows.
Buy once through Gumroad and use the included setup documentation to run locally.
RelayWorks AI Document Processing Kit
$79
FastAPI backend, Next.js frontend, local processing workflow, and AI-ready output packaging.
Final CTA
Start Building Instead of Integrating
Use RelayWorks when you need a working local document processing foundation for AI workflows without building every integration layer yourself.