Checklist and engines

Our Experience with Accessup.ai

Accessup.ai is a PDF remediation platform designed to help organizations bring their documents and e-books into compliance with the Americans with Disabilities Act (ADA) and the upcoming European Accessibility Act (EAA).

The platform integrates into high-volume PDF generation environments, allowing accessibility remediation to happen at scale.

Our Approach

We partnered with Accessup.ai to support the discovery phase and develop the alpha version of their product. With the EAA coming into effect in June 2025, Accessup.ai aimed to build a tool that could help automate document remediation with accuracy, scalability and compliance with existing accessibility standards.

Results

We delivered the discovery research and alpha build of Accessup.ai, including:

A clear understanding of legal and technical requirements for document accessibility

An analysis of current market offerings to identify gaps and opportunities

A validated approach for addressing two key use cases: PDFs and e-books

A functional alpha prototype ready for further development and testing

Visit Accessup.ai →

Implementation

We worked closely with Accessup.ai to define and prototype the initial version of their product. This included:

Understanding the regulatory and technical landscape — including ADA, EAA, and accessibility guidelines such as PDF/UA and WCAG

Market analysis — assessing existing tools and identifying key areas of differentiation

Discovery research — mapping workflows and needs for two core target formats: PDF documents and ebooks

Design and prototyping — turning ideas into functional prototypes to test core features

The Solution

The alpha version of Accessup.ai focused on a structured workflow that allows users to upload a PDF, apply remediation actions, and receive an updated version of the file along with a final accessibility report. The process combined established PDF processing tools with newer machine learning techniques.

1. Initial Report Generation

Each uploaded document is first validated using VeraPDF, which checks for compliance against an accessibility profile (e.g. PDF/UA-1). The output is an initial report highlighting areas that need attention.

2. Basic Accessibility Setup

Before applying remediation actions, the system checks and updates the document’s metadata and structure. This includes:

Tagging the document

Adding or correcting language information

Ensuring images have basic metadata

Including descriptions where possible

Documentation

3. Remediation Actions

The platform supports four key remediation actions:

Language Detection
An LLM reviews the document text and identifies the appropriate language, updating the metadata to reflect this.

Alternative Text Generation
All images are identified and tagged. Each image is then processed by an image-to-text model to generate a description, which is added to the document metadata.

Optical Character Recognition (OCR)
For scanned PDFs, OCR is applied to extract text from image-based content and insert it in a structured format with appropriate tagging.

PDF Layout Analysis (optional)
This tool provides a structural overview of the document to help verify logical reading order and layout consistency.

4. Final Report

After all selected actions have been completed, a final accessibility report is generated. This report reflects the updated state of the document, based on the remediations applied. See Accessup.ai in action with a hands-on preview here.

Technologies we used

Cloud Infrastructure — AWS

Programming Languages — Python

Tooling — Jupyter Notebooks, VeraPDF, OCR libraries

Machine Learning — PyTorch, large language models, image-to-text models