[Athen] Rec's for PDF remediation on a giant scale

[JORDISON_SHAWN via athen-list]

Kevin,

Those are some fantastic questions you have there.


  1.  In my experience, all of the math that we have tested comes through perfectly as MathML. I would argue that the OCR engine is better than ABBYY FineReader and is better than MathPix, both for the text and math respectively. It is treated as production-ready right out of the gate.
     *   Where there are charts and graphs, it will often write the alternate text in a meaningful way based on the content around the image. We have tested it with calculus-level math, but I haven't done as much testing around chemistry and biology specifically.
  2.  I think you bring up a good point about alternate text being meaningful and correct. What I would say is that Doc Access leverages multiple checkpoints for the alternate text. If for some reason something isn't accurate, it's very easy to jump in and edit in their workbench area for that specific file. Because it leverages AI, you can add customizations to the alt text for your campus in particular, and it will learn from the logic you apply to apply a similar methodology when generating alt text for future graphics.
     *   We have run this tool through the ringer with extremely complex files. While I will say it's not totally perfect, it's better than any tool I have ever seen.
     *   I would also advocate for a human check regardless, especially for scientific data.
     *   It doesn't simply slap the image into ChatGPT and provide alt text; it's more comprehensive than that. What I have noticed is that it's very factual about the image, but interpretive enough that it still makes sense for the user.
     *   One other component is that this can be customized based on where the tool is running (for example, in Canvas versus the public-facing content).
  3.  There can be a review process implemented automatically for pages that it flags as something that it had trouble understanding. It also gives you a breakdown of what exactly qualifies something to be flagged. When I say there is no faculty involvement, I mean that with the sheer volume of PDFs in courses, the faculty do not have to do anything specific for this tool to work its magic. They basically just upload files into Canvas like they normally do, and DocAccess handles all of the complex remediation from PDF to HTML without any user involvement. Now, if they want to jump in behind the scenes and make updates, they totally can, but it's not a requirement.

To be clear, DocAccess does not remediate PDFs, it provides an accessible HTML view (which can be downloaded too).


  *   Shawn

—
[cid:7399dec2113ef66672488768298bf1bb]
Shawn Jordison aka The Accessibility Guy - MS, Ed.D
<https://www.youtube.com/channel/UCPO0yzGQJmAa290DjrByhwQ>
Phone: 530-238-5645 or Book a  Free 15 minutes<http://www.tidycal.com/shawnjordison0000>
YouTube featuring Accessibility Tutorials<https://www.youtube.com/channel/UCPO0yzGQJmAa290DjrByhwQ>
<https://frontapp.com/signature>
[Sent from Front]
On February 6, 2026 at 9:56 AM PST ka791 at georgetown.edu<mailto:ka791 at georgetown.edu> wrote:


CAUTION: This email originated outside SMC.


Thanks, Shawn, that comparison between PREP and DocAccess is especially helpful. I’m genuinely curious to dig a bit deeper on DocAccess, because the claims around “fully automated” remediation are exactly what many of us are being asked about internally, often in the context of very large backlogs.


A few specifics I’d love more clarity on, if you’re willing to share from real-world use:

  *   MathML in practice: When DocAccess renders math via MathML automatically, how reliable is that output for complex equations and notation? Is there still a review step to confirm semantic accuracy and reading order with screen readers, or is it treated as production-ready out of the gate?

  *   Alt text and data tables: You mentioned detailed alt text and auto-generated data tables for charts and graphs. How are those validated for meaningfulness and correctness? In your experience, does this still require subject-matter review, especially for instructional or scientific content?

  *   “No staff involvement”: When you say faculty and staff don’t have to do anything, does that include QA and risk sign-off? Or is there still an implicit accessibility review happening somewhere in the workflow before content is considered compliant?


Asking from a place of due diligence rather than skepticism — the difference has big implications for cost, liability, and how we set expectations with leadership.


Really appreciate you sharing concrete experience here.

On Fri, Feb 6, 2026 at 12:50 PM JORDISON_SHAWN <JORDISON_SHAWN at smc.edu<mailto:JORDISON_SHAWN at smc.edu>> wrote:
Throwing in a +1 to DocAccess

This tool can render math accessibly via MathML automatically. It also has the ability to write detailed alternate text and will even provide data tables for charts and graphs that do not have a data table. There's also the ability to request a human check. It literally makes it to where faculty and staff do not have to do anything to the document. It's quite remarkable.

For the PREP tool I have always found that I still have to do some extra steps after the file gets out. In my opinion, it's not as automated as it seems, though it can handle simple files extremely well.


  *   Shawn

—
[cid:ii_19c3417b5b1fd9a4fbc3]
Shawn Jordison aka The Accessibility Guy - MS, Ed.D
<https://www.youtube.com/channel/UCPO0yzGQJmAa290DjrByhwQ>
Phone: 530-238-5645 or Book a  Free 15 minutes<http://www.tidycal.com/shawnjordison0000>
YouTube featuring Accessibility Tutorials<https://www.youtube.com/channel/UCPO0yzGQJmAa290DjrByhwQ>
<https://frontapp.com/signature>
[Sent from Front]
On February 6, 2026 at 9:46 AM PST athen-list at u.washington.edu<mailto:athen-list at u.washington.edu> wrote:


CAUTION: This email originated outside SMC.


Thanks for sharing this, Kamran — always helpful to hear what’s working in practice.


I’m curious about a few specifics, especially given the very strong claims PREP makes around compliance and scale:

  1.  STEM complexity: How does PREP handle documents with complex math, equations, and scientific figures in practice? In particular, what still requires manual human intervention versus what’s genuinely automated, and how is math accessibility (e.g., MathML, reading order, context) validated?

  2.  Alt text and figures: When PREP advertises “fully compliant” output, how is meaningful alt text handled for charts, diagrams, and images? Is that AI-generated, human-reviewed, or expected to be supplied by the institution? This has been a major cost and quality driver for us.

  3.  Verification and risk: What does your QA or validation process look like before documents are considered compliant (PDF/UA, WCAG 2.x)? Are you doing independent testing with assistive technologies, or relying primarily on automated checkers?

Asking because many of us are operating under significant budget constraints, and the difference between “accelerates first pass” and “reduces total human labor” really matters when making procurement decisions.


Appreciate any detail you’re willing to share.

On Fri, Feb 6, 2026 at 12:38 PM Kamran Rasul <krasul1 at jhu.edu<mailto:krasul1 at jhu.edu>> wrote:

Hi,



We have been using Prep by Continual Engine<https://www.continualengine.com/prep-pdf-remediation-software/>



It has helped us remediate large volumes of PDFs with short turnarounds.



[cid:ii_19c3417b5b04cff311]

Kamran Rasul, MEd.

Assistive Technology/Alternate Format Specialist (SDS)

Phone: 410-516-1167

E-mail: krasul1 at jhu.edu<mailto:krasul1 at jhu.edu>

Garland Hall, 1st Floor, Office 135-G

3400 N. Charles Street

Baltimore, MD 21218

Schedule a meeting with Kamran<https://outlook.office365.com/owa/calendar/ATC2@live.johnshopkins.edu/bookings/>



From: athen-list <athen-list-bounces at mailman22.u.washington.edu<mailto:athen-list-bounces at mailman22.u.washington.edu>> On Behalf Of Kevin Andrews via athen-list
Sent: Friday, February 6, 2026 12:02 PM
To: Monica Olsson <molsson at sbctc.edu<mailto:molsson at sbctc.edu>>; Access Technology Higher Education Network <athen-list at u.washington.edu<mailto:athen-list at u.washington.edu>>
Subject: Re: [Athen] Rec's for PDF remediation on a giant scale



      External Email - Use Caution





Jumping in with some lived-reality perspective from Georgetown, since we’re actively navigating this right now under pretty real austerity conditions.



We have an Acrobat Pro site license, and I want to be clear for folks reading along: that solves licensing, not remediation. At scale, Acrobat helps with OCR and basic tagging, but it does not eliminate the human labor required for structure, reading order, tables, math, figures, or meaningful alt text—especially for STEM content. That’s where time and money actually go.



Budget is a huge constraint for us at the moment, so the conversation internally has had to shift from “how do we make everything accessible” to “how do we make defensible, impact-driven decisions.” That’s meant being very explicit about tradeoffs:

  *   Not all legacy PDFs are equal in risk or usage.
  *   Automated tools are, at best, a first pass—not an end state.
  *   Vendor remediation often scopes out alt text, math, and complex figures unless you pay significantly more.
  *   Retrofitting thousands of documents is often more expensive than replacing or re-authoring the most critical ones over time.



What’s been most helpful for us is framing this as triage and lifecycle management, not a one-time cleanup project: prioritizing high-use / high-risk materials, de-emphasizing low-traffic archives, and being honest with stakeholders about what’s achievable under current financial conditions.



No silver bullets here—just hard constraints and deliberate choices. Sharing in case it’s useful context for others facing the same pressures.



On Thu, Feb 5, 2026 at 9:22 PM Monica Olsson via athen-list <athen-list at u.washington.edu<mailto:athen-list at u.washington.edu>> wrote:

Hi Amy, I plan to send you a document about Doc Access and security and thought that I had stated this in a prior email. Not a lot of folks know about the tool yet, in fact we (SBCTC) are they very reason a Canvas LTI tool even exists for the Doc Access tool. They built it for us :) I expect if our trial is successful and the tool is used at scale (really any scale) they will see a ton of higher ed clients over night…but we are kind of getting in at the beta stage. So, your inquiry to ATHEN may open a can of worms that I'm not entirely prepared for, but I guess it puts the tool on people's radar too!



Get Outlook for iOS<https://aka.ms/o0ukef>

________________________________

From: athen-list <athen-list-bounces at mailman22.u.washington.edu<mailto:athen-list-bounces at mailman22.u.washington.edu>> on behalf of Amy Rovner via athen-list <athen-list at u.washington.edu<mailto:athen-list at u.washington.edu>>
Sent: Thursday, February 5, 2026 5:46:49 PM
To: Wallace, Sagan <Sagan.Wallace at oregonstate.edu<mailto:Sagan.Wallace at oregonstate.edu>>; Access Technology Higher Education Network <athen-list at u.washington.edu<mailto:athen-list at u.washington.edu>>; Susan Kelmer <Susan.Kelmer at colorado.edu<mailto:Susan.Kelmer at colorado.edu>>; Joshua Hori <jhori at ucdavis.edu<mailto:jhori at ucdavis.edu>>
Subject: Re: [Athen] Rec's for PDF remediation on a giant scale



[Sent from outside SBCTC]



I'm very interested in the security of DocAccess. I hear that it can be embedded in Canvas. Is that how you are using it? I'm very suspect when companies using AI are too close to student data.



Thank you!

Amy



Amy Rovner, MPH RD
Director eLearning Services
Accessible IT Coordinator

Shoreline Community College

www.shoreline.edu<http://www.shoreline.edu/>  |  206.546.6937

eLearning Office: 206.546.6966

eLearning Email: elearning at shoreline.edu<mailto:elearning at shoreline.edu>

Shoreline Support Center<https://support.shoreline.edu/>



[Shoreline logo with tagline, Engage. Achieve.]

________________________________

From: athen-list <athen-list-bounces at mailman22.u.washington.edu<mailto:athen-list-bounces at mailman22.u.washington.edu>> on behalf of Joshua Hori via athen-list <athen-list at u.washington.edu<mailto:athen-list at u.washington.edu>>
Sent: Monday, February 2, 2026 2:56 PM
To: Wallace, Sagan <Sagan.Wallace at oregonstate.edu<mailto:Sagan.Wallace at oregonstate.edu>>; Access Technology Higher Education Network <athen-list at u.washington.edu<mailto:athen-list at u.washington.edu>>; Susan Kelmer <Susan.Kelmer at colorado.edu<mailto:Susan.Kelmer at colorado.edu>>
Subject: Re: [Athen] Rec's for PDF remediation on a giant scale



[ CAUTION: This email originated from outside Shoreline Community College. Do not click links or open attachments unless you recognize the sender and know the content is safe. ]



DocAccess has been looking very interesting. They not only use AI for alt tags but also integrate with AIRA.io to provide human audio descriptions of images.



Best,



Joshua



From: athen-list <athen-list-bounces at mailman22.u.washington.edu<mailto:athen-list-bounces at mailman22.u.washington.edu>> on behalf of Wallace, Sagan via athen-list <athen-list at u.washington.edu<mailto:athen-list at u.washington.edu>>
Date: Wednesday, January 28, 2026 at 8:16 AM
To: Access Technology Higher Education Network <athen-list at u.washington.edu<mailto:athen-list at u.washington.edu>>, Susan Kelmer <Susan.Kelmer at colorado.edu<mailto:Susan.Kelmer at colorado.edu>>
Subject: Re: [Athen] Rec's for PDF remediation on a giant scale

So far everyone I've looked at will do large-scale projects, but not write alt text (though they'll include it if you write it). I used Allyant for outsourcing PDFs at the time.



Since I'm feeling snarky....why not just dump them in an ABBYY hot folder and call it a day? They're not going to make accessible STEM PDFs anyway, so why waste time finding the best option? 🙃



Sagan Wallace

they/them

________________________________

From: athen-list <athen-list-bounces at mailman22.u.washington.edu<mailto:athen-list-bounces at mailman22.u.washington.edu>> on behalf of Susan Kelmer via athen-list <athen-list at u.washington.edu<mailto:athen-list at u.washington.edu>>
Sent: Wednesday, January 28, 2026 6:42 AM
To: Access Technology Higher Education Network <athen-list at u.washington.edu<mailto:athen-list at u.washington.edu>>
Subject: [Athen] Rec's for PDF remediation on a giant scale



[This email originated from outside of OSU. Use caution with links and attachments.]

I hope everyone's semester has gotten off to a bang-up start!



A department on my campus has a enormous stockpile of scientific/engineering-type PDFs that are publicly available, and that department wants to undertake turning them all into accessible documents.  Yes, they really want to do this, and first they asked me if I wanted to take on the project.



No thanks.



So I'm looking for recommendations for any large-scale PDF remediation companies that could do this kind of work.   Please and thanks.



Susan Kelmer

Alternate Format Production Program Manager

Disability Services

Division of Student Life

T 303 735 4836

www.colorado.edu/disabilityservices<http://www.colorado.edu/disabilityservices>



_______________________________________________
athen-list mailing list
athen-list at mailman22.u.washington.edu<mailto:athen-list at mailman22.u.washington.edu>
http://mailman22.u.washington.edu/mailman/listinfo/athen-list




--

Best Regards,

Kevin Andrews

Electronic and Information Technology Accessibility Coordinator

University Information Services
Georgetown University


--
Best Regards,
Kevin Andrews
Electronic and Information Technology Accessibility Coordinator
University Information Services
Georgetown University



--
Best Regards,
Kevin Andrews
Electronic and Information Technology Accessibility Coordinator
University Information Services
Georgetown University

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman22.u.washington.edu/pipermail/athen-list/attachments/20260206/cf822eba/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.png
Type: image/png
Size: 16012 bytes
Desc: image001.png
URL: <http://mailman22.u.washington.edu/pipermail/athen-list/attachments/20260206/cf822eba/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image002.jpg
Type: image/jpeg
Size: 28333 bytes
Desc: image002.jpg
URL: <http://mailman22.u.washington.edu/pipermail/athen-list/attachments/20260206/cf822eba/attachment.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: new-TAG-2025.png
Type: image/png
Size: 181742 bytes
Desc: new-TAG-2025.png
URL: <http://mailman22.u.washington.edu/pipermail/athen-list/attachments/20260206/cf822eba/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: new-TAG-2025.png
Type: image/png
Size: 181742 bytes
Desc: new-TAG-2025.png
URL: <http://mailman22.u.washington.edu/pipermail/athen-list/attachments/20260206/cf822eba/attachment-0002.png>