Executive Overview
In the modern digital landscape, web accessibility has transitioned from a niche technical consideration to a critical legal and ethical imperative. Yet, for small web agencies and independent freelancers, a persistent, uncomfortable operational bottleneck remains. When clients ask, "Is our website accessible?"—prompted by compliance fears, procurement questionnaires, or industry pressure—small shops face a painful economic reality. A rigorous, manual conformance audit requires days of specialist labor by certified Web Content Accessibility Guidelines (WCAG) experts. At small-shop billing rates, this level of deep analysis is rarely financially viable.
Consequently, the question often lingers unanswered or receives a vague, generalized brush-off that satisfies no one and leaves clients exposed to legal liabilities and user alienation.
Enter Sheree Preston, founder of Curbcut. Faced with this recurring industry pain point, Preston set out to build a pragmatic solution: a free, white-label accessibility scanner designed specifically for resource-constrained agencies. Leveraging a collaborative team of AI agents—an AI CTO and an AI engineer—Preston engineered, tested, and deployed the core application in approximately one week.
Curbcut does not claim to replace a multi-day manual audit by a certified accessibility professional. Instead, it offers a fast, plain-English first look. By analyzing a URL, the tool generates a scored, prioritized report detailing broken elements, specific WCAG 2.2 violations, impacted user demographics, and actionable code fixes. Crucially, the platform features a white-label mechanism, allowing agencies to brand the diagnostics as their own deliverables. This transforms an awkward compliance question into an immediate business conversation and an entry point for lucrative remediation work.
This report examines the genesis of Curbcut, the unique paradigm of human-AI collaboration that brought it to life in seven days, the technical hurdles overcome during development, and the vital philosophy of radical transparency regarding what automated tools can—and fundamentally cannot—achieve.
Detailed Chronology: A Week-Long Sprint with an AI Development Team
The development of Curbcut challenges traditional software engineering timelines. Rather than spending months assembling a team of developers, Preston acted as the visionary product director, managing a specialized workforce of autonomous AI agents.
Day 1–2: Conceptualization and Core Architecture
The project began with a clear problem definition: small agencies need an accessible, low-friction entry point to digital compliance scanning. Preston outlined the product parameters in plain-English specifications, which were then processed by an AI CTO. The AI CTO broke down the product requirements into structured development epics, delegating concrete execution tasks to an AI engineer agent.
During this initial phase, the core scanning pipeline was established. The goal was simple: ingest a URL, parse the Document Object Model (DOM), evaluate elements against WCAG 2.2 standards, and synthesize the findings into readable, actionable insights.
Day 3–4: Iterative Failure and Edge-Case Discovery
Software development is rarely linear, and human-AI collaboration relies heavily on trial and error. Two pivotal episodes during the build phase defined Curbcut’s current feature set:
-
The Staging Site Dilemma: Early in internal testing, Preston accidentally pasted an email address into the URL input field instead of a web address. The system returned a confusing, generic error. Upon sending a screenshot of the failure to the AI team, the interface was rapidly updated. However, chasing the root cause of this error uncovered a massive industry blind spot: standard accessibility tools are notoriously incapable of scanning password-protected staging environments—which is precisely where agency development work takes place.
Within days, the AI team engineered a secure solution: support for staging links that embed sign-in credentials directly within the URL string. To mitigate security risks, strict guardrails were implemented. Credentials are transmitted exclusively to the exact host specified by the user and are permanently scrubbed from final reports, system logs, and cache layers. What began as a user input error transformed into one of Curbcut’s most distinctive enterprise-grade features.
-
Navigating Firewall Barriers: Another early test involved a website protected by an aggressive Web Application Firewall (WAF) that automatically blocked automated scraping and scanning agents. The initial scanner failed, returning an ambiguous status code. The AI team refined the diagnostic messaging to clearly state that the target website’s firewall had rejected the automated visitor, helping users distinguish between a functional software bug and an external security block.
Day 5–7: Testing, Verification, and the "All-Green" Reality Check
As the feature set stabilized, the AI engineering agent executed rigorous automated test suites to verify system reliability. Preston coupled this with manual, human-in-the-loop sweeps of live websites.
A critical psychological milestone occurred when the tool produced its first flawless, "all-green" accessibility report on a test site. Rather than celebrating prematurely, Preston asked a more critical question: "What did the tool fail to check?"
That skepticism reshaped the product’s architecture, reinforcing the boundaries of automation and cementing the platform’s commitment to radical honesty regarding the limitations of programmatic code analysis.
Supporting Context & Metrics: The Mechanics of Curbcut
To understand Curbcut’s utility, it is necessary to examine how it operates within an agency’s workflow. The platform is engineered to minimize friction and maximize commercial value for web professionals.
Core Functional Capabilities
- Instant URL Ingestion: Users paste any publicly accessible web address (or authenticated staging link) into the interface with zero required software installation or account registration.
- Plain-English Diagnostics: Complex technical violations are translated into clear, human-readable explanations that demystify accessibility compliance.
- WCAG 2.2 Alignment: Identified issues are mapped directly to official WCAG 2.2 success criteria, ensuring industry-standard accountability.
- Impact Mapping: Reports explicitly outline which user groups are affected by specific accessibility barriers (e.g., individuals with low vision, motor impairments, or cognitive disabilities).
- Actionable Remediation: Alongside problem identification, the tool provides suggested code fixes, equipping developers with the exact syntax needed to resolve errors.
- White-Label Agency Branding: Agencies can input their own company name alongside the client’s name, generating a polished report titled "Prepared for [Client] by [Agency]." This seamlessly positions the agency as a proactive digital partner.
The Economic Engine for Small Shops
For a small digital agency, a comprehensive accessibility audit can cost thousands of dollars—a price tag that prices out smaller clients who need help the most. Curbcut bridges this gap. By offering a rapid, zero-cost initial diagnostic, agencies can open a constructive dialogue with clients. Once the report highlights critical vulnerabilities, the conversation naturally shifts toward remediation and development retainers—the very services that drive agency revenue.
Official Statements and Developer Insights
Reflecting on the unprecedented speed of the build and the philosophy driving the platform, founder Sheree Preston emphasizes the necessity of honesty in the automated compliance space:
"If you run a small web agency, or freelance for a living, you know this email. A client has read something, a procurement form asked, or a competitor got called out—and it lands in your inbox: ‘Quick question—is our website accessible?’ There’s no quick answer. The real one is a conformance audit—specialist work, days of it, by someone who does WCAG for a living. That’s expertise a small shop can rarely justify."
Detailing the collaborative dynamic with her AI development team, Preston notes the distinct division of labor between human intuition and machine execution:
"No mysticism: I didn’t write the code, and I’m not an accessibility specialist. What I have is a team of AI agents—an AI CTO who turns my direction into a plan and delegates concrete tasks, and an AI engineer who implements, tests and verifies. My job is the part that can’t be delegated: describing the product in plain words, making the calls, and judging results the way a client would."
Crucially, Preston remains intensely focused on maintaining user trust by actively publicizing what the tool cannot do:
"This is the part I care about most, because overclaiming is how trust dies. When I got my first all-green report, I didn’t celebrate—I asked what the tool had not checked. The answer changed the product again."
Future Outlook and What Curbcut Cannot Do
In the software industry, overpromising capabilities is a chronic hazard—particularly within the automated accessibility market, where automated "overlay" widgets have historically faced severe backlash from disability advocacy groups. Curbcut deliberately rejects this deceptive paradigm by embracing radical transparency regarding its limitations.
Explicit Limitations of Automated Scanning
Automated tools, regardless of sophistication, operate within strict technical boundaries. Curbcut openly acknowledges that it cannot:
- Validate Contextual Semantics: While code can be checked for missing
alttext, an automated scanner cannot determine if the provided alt text is actually descriptive and meaningful in the context of the surrounding content. - Evaluate Keyboard Navigation Logic: Complex interactive components—such as custom modal dialogs, dropdown menus, and dynamic single-page application (SPA) state changes—require human tactile testing to ensure screen readers and keyboard-only users can navigate them intuitively.
- Certify Legal Compliance: No automated software can legally guarantee compliance with the Americans with Disabilities Act (ADA), Section 508, the European Accessibility Act (EAA), or WCAG conformance levels. Only comprehensive human audits combined with user testing involving individuals with disabilities can achieve true verification.
The Road Ahead: Real-World Testing and Iteration
As Curbcut enters its public testing phase, the platform remains entirely free while its long-term economic model is evaluated. Significantly, the product’s evolution thus far has been driven entirely by internal testing, developer mistakes, and automated sweeps.
The immediate future of the platform will not be dictated by rigid roadmap theatre or speculative feature bloat. Instead, future updates will be shaped organically by real-world agency users scanning production and staging sites.
Web professionals, independent developers, and digital agency owners are encouraged to put the platform through its paces—testing their most structurally complex or deliberately flawed websites. By contrasting the diagnostic outputs against real-world client expectations, the broader web development community will help determine whether AI-assisted triage can permanently solve one of the industry’s most persistent operational challenges.
Curbcut is currently available for public use at no cost. Agencies and developers can evaluate the tool directly by visiting the live application environment.
