Inspiration

What does agreeing mean if the person agreeing does not understand?

Vera ScholarA⁺ began with that question.

We encounter terms, conditions, permissions, and obligations throughout our digital lives. Before we can enroll, access a service, submit information, or activate software, we often face language that requires more attention and background knowledge than the surrounding experience prepares us to give.

An institution can make information available. Our question was how to help make that information understandable at the moment a person needs it.

At Broward College’s AI and cybersecurity hackathon, we brought together an interest in philosophy, ethics, and public comprehension with an interest in programming and web development. Bryan Warfield approached the problem through the relationship between understanding and individual choice. Arlen Hernandez contributed programming knowledge and an HTML structure that helped shape the project.

We started with handwritten notes and a broad idea about digital agreements. Through discussion and iteration, we narrowed the first application to a situation we could demonstrate clearly: a student trying to understand a college agreement.

That became Vera ScholarA⁺.

What it does

Vera ScholarA⁺ is a prototype for connecting institutional agreements with accessible explanations inside the user’s existing experience.

The central interaction is a toggle. A student can switch between the original wording and an enhanced, plain-language explanation without leaving the section they are reading. Switching back restores the exact original text.

Our demonstration covers three fictional situations:

  • Withdrawal and attendance requirements.
  • A student portal provider’s access to submitted files.
  • The difference between estimated financial assistance and a guaranteed payment.

For example, a student may encounter:

“Nonattendance does not constitute withdrawal.”

The enhanced explanation makes the practical distinction clear: stopping attendance does not automatically withdraw the student. The full demonstration also preserves the requirement to submit a withdrawal form and the conditions surrounding refunds.

The purpose is to help someone understand the decision in front of them. The original agreement remains accessible, and clarification guidance points toward the relevant institutional office when the example does not contain an answer.

The current prototype uses prewritten explanations. It does not connect to a live AI model, process real student records, or record acceptance of an agreement.

The connection we want to build

Our vision is a connection between existing institutional systems, approved source material, and the people expected to act on that material.

Institutions already have agreements and procedures. Users already encounter them. Vera ScholarA⁺ would introduce an explanation layer at those existing points of interaction.

In a future implementation, an institution could connect its current documents to an AI-assisted drafting process. Authorized staff would review the explanations before publishing them alongside the corresponding source text.

This approach gives institutions a way to improve communication while preserving responsibility for the language they publish. It also gives users a way to examine unfamiliar wording before deciding whether to proceed.

We want transparency to be something people can use.

How we built it

We built the demonstration with HTML, CSS, and JavaScript:

  • HTML contains the page structure, original clauses, enhanced explanations, and switches.
  • CSS controls typography, spacing, responsive layouts, colors, and the appearance of each switch.
  • JavaScript listens for changes and determines which text version appears.

The interaction follows a straightforward algorithm:

  1. Load the agreement sections with their original wording visible.
  2. Loop through the sections and locate each switch and its two text versions.
  3. Attach a change-event listener to each switch.
  4. When a switch turns on, hide the original paragraph and reveal the enhanced explanation.
  5. When it turns off, restore the original paragraph.
  6. Keep each section independent so the user controls their reading experience.

We produced both a self-contained HTML demonstration and a version with separate source files. That separation matters to us: we wanted to explain how the product works, not simply show a finished interface.

The visual design evolved into a deep navy background with bright, readable text and a subtle silver treatment on the branding. We also included keyboard-operable switches and a layout that adapts to smaller screens.

How we used AI

AI supported the development process from the beginning. We used ChatGPT to organize the idea, explore its boundaries, draft code, explain programming decisions, and revise the experience through successive prompts.

Our prompts became more specific as our thinking developed:

  • We moved from general terms and conditions to college agreements.
  • We requested a start-to-finish algorithm.
  • We replaced buttons with switches between original and enhanced text.
  • We separated HTML, CSS, and JavaScript so the implementation could be examined.
  • We refined the typography and color scheme to establish a clearer identity.

Each request translated an intention into a concrete design or coding decision.

One important distinction emerged: using AI to build the prototype is different from connecting AI to the running product. Our demonstration shows the user experience. Live document processing and institutional review remain future development work.

A path to a serverless product

The current demonstration runs entirely in the browser. It can open as a local file and does not require an application backend.

That provides a lightweight starting point for a future serverless architecture.

The student-facing interface could run on static hosting, while managed backend functions handle tasks such as receiving an authorized document, requesting an AI-generated draft, and publishing reviewed content. “Serverless” means a cloud provider manages the underlying server infrastructure; it does not mean that online services use no servers.

Our proposed workflow is:

  1. An authorized institution employee submits an agreement and its version.
  2. A backend function requests draft explanations grounded in that document.
  3. The system stores the drafts separately from published content.
  4. A designated reviewer checks meaning, qualifications, and exceptions.
  5. Approved explanations become available beside the matching agreement.
  6. A document update triggers another review of affected explanations.

The user’s toggle would retrieve or display an approved explanation. It would not need a new AI request every time someone switched views.

That distinction could help control operating costs and keep explanations consistent. Actual cost and performance would still depend on usage, document size, hosting choices, and the selected AI service.

A production version would also need authentication, access controls, secure handling of API credentials, version management, and appropriate retention rules. Serverless infrastructure would support those responsibilities rather than remove them.

Challenges we faced

Our first challenge was scope. The initial idea could apply to almost any digital agreement. For a short hackathon, we needed a focused example that made the interaction understandable and achievable.

Our second challenge was preserving meaning. Shorter wording is not automatically a faithful explanation. Removing an exception or condition could change what a person thinks they are agreeing to. That led us to preserve the original text and make institutional review part of the proposed production workflow.

We also had to distinguish clarity from acceptance. Viewing an explanation does not prove understanding, and flipping a switch does not constitute agreement. Our prototype keeps those actions separate.

Another challenge was connecting the concept honestly to cybersecurity. We focused on understanding permissions and third-party access to information. The prototype explains a fictional data-access clause; it does not detect malicious software or determine whether a service is safe.

Finally, we had to make the code understandable to ourselves. Asking for comments, separated files, and an explicit algorithm helped turn an AI-assisted output into something we could inspect, discuss, and continue developing.

What we learned

We learned that a small interaction can express a much larger idea.

The toggle is simple, but it represents a deliberate choice: the user should be able to move between institutional wording and an explanation without losing access to either.

We also learned that useful AI development requires specific direction. The quality of the experience improved as we clarified the audience, narrowed the problem, and explained exactly what we wanted the code to do.

Most importantly, we learned that transparency needs both accurate information and an accessible way to engage with it. Whether our approach improves comprehension is something we intend to test with users.

Expansion beyond education

Education is our starting point. We see several possible applications for the same approach.

Veterans Affairs and veteran services

As a Navy veteran, Bryan sees a meaningful potential application in helping veterans navigate unfamiliar administrative language.

A future version could explain application instructions, documentation requirements, information-sharing permissions, and procedural obligations alongside official materials. Agency or service-provider reviewers would need to verify those explanations.

The goal would be to help veterans understand the process and identify the right place to ask questions. The software would not determine eligibility or promise a benefits outcome.

Health insurance

Health insurance presents another possible setting for contextual explanations.

A future integration could help members understand terminology in their own plan documents, including deductibles, coinsurance, exclusions, prior authorization, and network restrictions.

Explanations would need to reflect the specific plan, its exceptions, and its current version. Any use of personal health information would require additional privacy and security review. Our present prototype does not handle that information.

Software modernization

Organizations could potentially add the explanation layer to existing customer portals, onboarding flows, and permission screens.

This would allow the concept to support software people already use. An organization could explain what a permission authorizes or what an agreement requires within the existing workflow.

These are proposed expansion opportunities. We have not established partnerships or validated demand in these sectors.

What comes next

Our next step is to test the experience with students and institutional staff.

We want to learn whether users interpret a clause more accurately after viewing its explanation, which wording still creates confusion, and how much effort institutions would need to review and maintain the content.

From there, we would develop the review dashboard, connect a controlled AI-drafting workflow, and test a serverless deployment with versioned agreements.

For the presentation, we prepared a PowerPoint deck and demonstration assets, with DaVinci Resolve planned for video editing and Devpost as the submission platform.

Our ambition is to help institutions make understanding part of the experience they already provide.

Vera ScholarA⁺: Clarity before commitment. Understanding before action.

Built With

Share this project:

Updates

Submission history