Free for Government and Universities

Audit, triage, and remediate PDFs

ASAP PDF AI helps agencies discover, audit, prioritize, and remediate PDFs across their public websites.

This tool combines traditional ML (document classification) and LLMs (summaries, exception checks) to reduce the work needed for ADA/WCAG compliance.

Request Free Access
Why this matters now

By April 26, 2026, ADA-compliant government websites must meet WCAG 2.1 AA accessibility standards, and PDFs are often the final remaining barrier to full compliance.

Proven pilots

Code for America launched pilots with Salt Lake City and the state of Georgia. We support onboarding for other governments.

Technical overview

ASAP PDF AI combines three main layers: discovery, classification & extraction, and AI-assisted decisioning.

Discovery

A web crawler pulls public PDFs and metadata (URL, title, dates, publisher) so teams can understand scope and ownership.

Classification & extraction

Lightweight ML classifies documents by type using URL, filename, and surrounding text. The system extracts text from scanned and born-digital PDFs for search and summaries.

LLM-assisted decisioning

LLMs generate summaries, suggested alt text, and an exception-checker that compares a PDF’s content to ADA exception criteria to recommend archival or remediation.

What the AI PDF Audit Tool does

A purpose-built workflow for state and local government teams to gain visibility, make informed decisions, and accelerate compliance.

Automated discovery

Web crawling locates and inventories every public PDF across your domains, capturing metadata like title, creation date, author, and URL.

AI classification

Machine learning categorizes documents (e.g., forms, agendas, maps, brochures, policies) to support bulk prioritization and routing.

Intelligent summaries

Large language models generate concise, human-readable summaries and draft alt text for images—accelerating review.

Exception checking

Comparisons against policy criteria help reviewers quickly determine whether a PDF may qualify for an exception or needs follow-up.

Smart search & filters

Search by filename, type, complexity, and review status to surface quick wins and assign complex items to experts.

 

Progress analytics

Dashboards show inventory size, decisions made, remediation throughput, and remaining risk—ideal for leadership reporting.

Built for government teams

Purpose-built workflows

Designed with innovation officers, content managers, and digital service strategists to fit existing review and publishing processes.

Scalable and cost-effective

Reduce manual effort and triage faster across large PDF inventories. Focus people on the highest-impact work.

Secure and low-risk

Operates on publicly available documents and supports open-source deployment patterns common to public sector IT.

What you can do with it

Program leads
  • See current risk and forecast remaining work
  • Direct resources to high-impact categories (e.g., forms)
  • Report progress to leadership and the public
Content teams
  • Prioritize “simple” PDFs for quick wins
  • Use summaries and draft alt text to accelerate review
  • Apply policy exceptions with confidence and consistency

  • Comprehensive discovery – a web crawler gathers every public PDF on your domain so you can see scale and scope at a glance.

  • Smart classification – document types (forms, agendas, maps, brochures) are auto classified using lightweight ML to help you prioritize. (Code for America reports ~81% accuracy for their classifier in early models.)

  • LLM-powered summaries & alt text – generate short summaries and suggested alt text to speed reviewer decisions on large or image-heavy PDFs.

  • AI exception checker – compares a document to ADA exception criteria and surfaces likely archive/remove cases to reduce unnecessary remediation.

  • Prioritized remediation guidance – targeted recommendations (tagging structure, reading order, tables, headings) so staff can focus on what matters.

  • Analytics dashboard – track review progress, scope, and outcomes across departments.

Request Free access

Complete this intake and our team will contact you to assess and map an onboarding plan.







FAQ

Is this truly free?

Yes – Code for America’s ASAP PDF AI project is being made available to all government agencies at no cost.

How long does onboarding take?

Intake, discovery crawl, and an initial audit typically take 1–2 weeks depending on size. We provide a pilot plan and timelines during intake.

Can we trust LLM outputs?

 LLM outputs (summaries, alt text, exception suggestions) are meant to assist reviewers, not replace human judgment. The system includes evaluation tooling and human-in-the-loop review workflows.

Will this make us ADA/WCAG compliant?

ASAP PDF AI automates much of the heavy lifting (audit and remediation guidance) but final compliance decisions and sign-offs rest with the agency. The tool reduces effort and risk significantly.

Does the tool edit PDFs?

The tool focuses on discovery, triage, and decision support. Remediation and conversion can be handled by existing tools and vendors, informed by the audit.

Can it handle very large PDFs?

Yes. LLM-based summarization is designed to reduce long documents into actionable overviews for reviewers.

How accurate is classification?

The ML classifier is tuned for government document types and evaluated for reliability; reviewers remain in the loop for final decisions.

Next steps

Ready to try ASAP PDF AI? Request access via the form, or email us at asap@keyspider.io with: agency name, contact, and estimated PDF scope. We’ll follow up to schedule an intake and demo.

Ready to tackle your PDF backlog?

 

Start your audit now and accelerate toward ADA compliance before April 26, 2026.

Request access
Read Code for America blog