Scaling the Visual Experience Platform

First published on the Barnes Foundation's Medium, August 2026.

How we took a platform built for the Barnes method and opened it to the field

In a previous post, I walked through the research, design, and 2023 launch of the Barnes Visual Experience Platform, and the structured user research process that produced Replay, our asynchronous playback feature, in March 2024, which opened up evergreen self-paced content both for courses built for on demand and for recordings of live classes that stay explorable afterward. I’m picking up where we left off to cover what happened when another institution started teaching on the platform, the Admin 2.0 research that reshaped the next version, and what’s new this fall.

A 3D model of a kylix on a VXP board in Edit mode. Instructors and students can rotate the object and move in close, the way a visitor circles a case in a gallery, then go further and look inside the bowl at the gorgoneion, which no case would let you see. Eye Cup, c. 530 BCE, Greek, Attica, terracotta.

The Partnership

In July 2025 the Barnes announced its partnership with the Penn Museum, the first external institution to license the VXP. Beginning in fall 2025, the Penn Museum hosted its Deep Dig classes and its Archaeology in Action lecture series on the platform. This was the first test of the licensing model which we’d been turning over as a “what if” since peers started asking questions after we presented the VXP at conferences like MCN and AAM. A platform built to support the Barnes pedagogy and mission is now also a revenue line and in service to the field (and to some of our closest neighbors here in Philadelphia!)

Penn Museum’s programs use visual material differently than the Barnes adult education catalog: archaeological objects, site photography, mural exhibitions, and material culture rather than paintings and ensembles. But the interaction model we built for the Barnes method, shared attention with bounded individual exploration, was a natural fit.

The partnership also changed what the platform is to the institution. Licensing revenue helps fund the VXP’s continued development, and it pushed us to build the pricing, support, onboarding, and documentation a partner needs to run their own programs. A licensing model that sustains further development on the platform and investment in our own programs and mission is what makes that access commitment possible.

Since 2023, over 7,000 students and event attendees have learned on the VXP. One in five has been a scholarship recipient or an Art for All Community Pass holder. 82% percent of students say taking a course on the VXP makes them more likely to take another course on the VXP, 91% say it inspired them to continue their learning, and 80% share that the VXP had a high or extremely high impact on the ways in which they think critically about art. Instructor feedback has been glowing. 100% report that the VXP had a positive impact on student engagement and 90% that it positively impacted their teaching.

A licensing model that sustains further development on the platform and investment in our own programs and mission is what makes that access commitment possible.

After year one of serving another institution on the platform, the model still depended on manual work from the Barnes team. Every class required some hands-on coordination on content prep, board building, recording, and incident response. The partnership proved the platform could host another institution’s pedagogy. It also proved the platform could not yet host another institution’s operations.

That gap defined the next research and development cycle, and what we’re calling VXP 2.0.

*An instructor edits a board in VXP 2.0: searching the Barnes ensemble images for Room 4, placing the south wall beside the object, then zooming in to find the same dish on the gallery wall.*

“Admin 2.0”: Research Before the Rebuild

I came to software engineering and product management from design thinking and human-centered design, and those instincts really came into play here. Every project starts with these two questions. What problem are we solving? And is this the right problem to be solving right now?

We ran this playbook with the development of our asynchronous Replay offering: Define the problem with structured user research, let the findings build the backlog and drive the strategy. Replay was student-facing. This round was about the part of the platform students never see. The admin side had to work before we could take on more partnerships or move anyone toward self-serve, so we scoped this slice of the rebuild as “Admin 2.0”.

We wanted to answer: what does the platform have to become for an external partner or an instructor to teach on it confidently, without Barnes staff in the loop? Confidently is a higher bar than it sounds, and all instructors don’t teach in the same style. Some build every board from scratch, some adapt last semester’s, some arrive with a conference talk or lecture materials from another institution already in PowerPoint. A platform that only accommodates one of those sends everyone else looking for a workaround.

Two things had to stay true in whatever we built. Every feature we added or improved had to serve both our own education programs and a partner’s programs, because we weren’t going to run two roadmaps. And each new partner had to cost less to onboard than the one before. We can staff the jump from one partner to two. Getting from two to ten meant driving the activation energy down until a partner could start without us. So we measured scoping decisions by how much Barnes labor would be left over for a partner after we shipped.

Two research questions framed the work:

What are the current board-building workflows, pain points, and opportunities for improvement?
Which feature updates are high effort / high impact, high effort / low impact, low effort / high impact, low effort / low impact?

Across 15 interviews comprising of both internal and external users of the VXP, we used a mixed-methods approach: a quantitative importance-rating exercise scored against internal effort estimates, qualitative thematic coding of interview commentary, and user stories drawn from participants to articulate specific challenges in their own framing. The output was an effort-by-impact matrix and a prioritized list of recommended investment areas, so that staff can weigh impact against effort on their own and get more deliberate about when the answer is yes, no, or not yet.

100% of instructors said the VXP had a positive impact on student engagement and 90% said it positively impacted their teaching.

The strongest and most consistent finding across every interview was that the student-facing platform works. Live teaching is strong. Replay is working. Engagement tools are working. That tracked with our post-course surveys. The friction is not in the classroom, rather the friction is upstream. The platform succeeds once content exists. The primary failure point for instructors on VXP 1.0 is content creation, reuse, and management.

1. Presentation creation improvements (highest priority). “Every class that I teach is in the service of a future class that I will teach because I don’t want to redo my work.” Instructors expect to start from existing material, whether that’s last week’s slides, last semester’s class, or a presentation from a conference or teaching somewhere else, and adapt. VXP 1.0 forced full rebuilds. The recommendation set covered drag-and-drop reordering, insert-at-position, batch upload with auto-sequencing, dedicated text board templates to eliminate the PowerPoint-screenshot-to-JPEG workaround, and a PowerPoint import path that treats PPT as a supported planning on-ramp rather than a competing workflow.

2. Guided onboarding flows. “I don’t really know where to start.” Instructors lack a clear mental model of what a VXP presentation is and how to begin. The recommendation was a guided Create Presentation entry point with intent-based starting paths, meaning reuse existing, start blank, import from PowerPoint, or start from template, paired with a Create Course flow with class-type presets for live, async, hybrid, and in-gallery use.

3. Asset management improvements. “Even if I know we’ve used an image before, I usually just re-upload it.” Image discovery and asset management surfaced as a major time sink across every interview. The recommendation was a centralized Asset Library with stronger search refinement, duplicate detection, resolution indicators, usage history, and safe cleanup workflows that distinguish reusable assets from replay-dependent legacy content.

The matrix also identified what not to build, at least for now. Member portal integration, and support for class sizes above 250 students all rated lower on instructor impact than the authoring work. They remain on the roadmap as institutional or strategic investments, but they are not on the critical path to self-serve.

Penn Museum proved a partner with existing, impactful programming could teach on the VXP. The Admin research defined what the platform has to become for a partner to operate on it.

What Shipped

The last class on VXP 1.0 met on August 19. We took the platform down for two days, repointed the URL at the new client, and fall classes start on 2.0 on September 9.

This might sound like a hard cutover but in reality it wasn’t. VXP 2.0 ran against the same server and the same production database as 1.0 all summer, so both interfaces could be live at once and an instructor could join or manage the same meeting from either one. That’s what made iterative testing possible on real courses, real images, and real accounts instead of on a staging environment. It also meant we never had to migrate everyone in one night. We piloted instructor testing on the new platform through July and August while classes kept running on 1.0, and by the time we retired the old client everyone on our internal education team had the opportunity to test and provide feedback on 2.0.

The rebuild concentrated on the client and the operational layer. The front end now runs on a modern App Router framework (Next.js) and the shared Mantine design system we use across Barnes and Calder Gardens digital properties. State management and data fetching were refactored and the real-time canvas event infrastructure was rebuilt for scale and reliability. The Admin Panel was redesigned from scratch, grounded in the research above.

The new and improved course dashboard to strengthen self-serve capabilities. *Manage courses and sharing in one place. View status and reprocessing without support.*

What that means for an instructor building a fall class: guided course and presentation “Getting Started” wizards, low-awaited board building and content production enhancements, rich text captions on every board, a course dashboard with search, filters, and a calendar view, recording management with session status at a glance, and organization-scoped member management and roles. More robust asset management, PowerPoint import and self-serve 3D, panorama, and text boards are coming soon (stay tuned!).

Starting a presentation. Reuse an existing one, start blank, import from PowerPoint, or start from a template, which answers the two things instructors told us most often: I don’t know where to start, and I don’t want to redo my work.

We took this maintenance time as an opportunity to ship some impactful accessibility work alongside the admin improvements. A student can pan and zoom a deep-zoom image with keyboard alone, which matters more than it sounds for a platform whose entire premise is close looking. Screen readers announce the image, panorama, and 3D viewers, and read out each new board without interrupting the class. Live closed captions run in class and caption tracks run on every replay. Focus indicators are visible for keyboard users, animations respect reduced-motion settings, and touch targets meet WCAG 2.5.8 across mobile and tablet. A full audit and remediation pass is scheduled for later this year.

Board building improvements in response to user feedback. *More intuitive, PowerPoint-like experience so instructors and admins can self serve build their own boards.*

Writing the Manual Inside the Product

The multi-tenant boundary was already in the database. Penn Museum has operated against it since 2025. The remaining work was the admin interface that empowers a partner, and our own instructor teams, to use the platform without our involvement.

So 2.0 ships with documentation in the platform rather than in a folder or document attached to an email somewhere. There’s a Guides section in the admin panel covering FAQ and platform onboarding: how deep zoom works, what IIIF is doing behind the scenes, what resolution to send us, and where to find high-resolution images in the first place. We’re adding a native changelog so users can see what changed and when.

*Written documentation for admins in the admin panel and improved step-by-step help for students inside every session.*

There’s also help inside the live session. A student can learn to navigate the board without leaving class, and the panel explaining Focus and Explore modes describes the Barnes method. The student lobby got this treatment too. It used to be a waiting room where you checked your camera and mic. We added a deep zoom canvas with a collection work to it, so a student who arrives early can chat, sort out device settings, and warm up their eyes on a close looking exercise instead of watching a countdown. The pedagogy is in the interface!

What’s Next

Fall classes start September 9. The rest of the roadmap follows shortly and is arguably just as impactful as we continue to scale: PowerPoint import, a robust asset management library and system, self-serve 3D and panorama and text boards.

Museums, universities, and cultural organizations can license the VXP to run their own visual teaching programs, with their own collections, speakers, and course calendars. A partner organization manages its own courses, sessions, members, and roles inside the platform, and we handle onboarding, technical support, and production help through the first term. What comes with it is a century of practice in teaching people how to look, and the Barnes team to help your educators put it to work.

If any of this maps onto work happening at your institution, I’d love to compare notes. The VXP roadmap has already been shaped by people outside the Barnes, and there’s room for more! Please get in touch.

Experience the VXP in this preview lesson, then take a live class or a self-paced on-demand course as a student. Museums, universities, and cultural organizations can license the VXP to run their own visual teaching programs, using their own collections, speakers, and course calendars. For institutional inquiries about the platform and partnerships, contact us. Follow the Barnes Foundation on Medium for the rest of the series. If you are just joining, start here, then read how we designed and built the platform.

Scaling the Visual Experience Platform was originally published in Barnes Foundation on Medium, where people are continuing the conversation by highlighting and responding to this story.