Want to know more?
Are you interested in this project? Or do you have one just like it? Get in touch. We’d love to tell you more about it.
How a two-person team delivered a production AI audit platform in six weeks
Our client PX set out to reduce compliance risk across a fast-growing marketplace of lead generation partner websites. Operating in a regulated environment, they needed a reliable way to ensure that hundreds of third-party sites consistently displayed the required legal disclosures, without relying on manual review.
Equal Experts worked alongside the team at PX to design and deliver an AI-powered compliance audit platform that could autonomously navigate partner websites, assess disclosure requirements and report findings in a consistent, repeatable way.
Together, the teams combined PX’s deep expertise in lead generation and compliance with Equal Experts’ experience in AI-assisted product delivery. Using our Dialogue-Driven Delivery framework, a two-person team built and deployed a production system in just six weeks.
The result was a scalable compliance capability that reduces operational risk, supports future growth and enables PX to protect both its business and its customers.
Outcomes
17%
average sales increase in first month
3 months
from concept to launch
6 weeks
to production deployment
About PX
PX is a customer acquisition platform connecting leading direct-to-consumer advertisers with a marketplace of over 1,200 lead generation partner websites. Known for its performance-driven approach and strong compliance standards, PX operates in a highly regulated e-commerce space where transparency and responsible data practices are essential to long-term success.
Industry
Marketing and Technology
Organisation size
1,200+ partner websites in its marketplace ecosystem
Location
Offices in the Netherlands, New York and Panama
Project length
6 weeks
Challenge
Scaling compliance oversight across hundreds of evolving partner websites
PX works with a large network of publishers generating leads in a range of markets. Each partner website must display specific legal disclosures, such as TCPA (Telephone Consumer Protection Act) consent language, up-to-date privacy policies, terms and conditions and partner lists.
Historically, compliance checks were manual. Team members had to navigate individual websites, click through funnels, search for disclosure text, and log the results. This created several challenges:
- High operational effort
- Risk exposure if changes went unnoticed
- Limited scalability as the network grew
- Dependency on manual processes in a fast-moving environment
The need was urgent. Mark van Heyningen, Chief of Staff at PX explains:
“It’s about risk reduction for our clients. There’s a lot of manual work, and publishers frequently change their websites. We needed a better way to stay on top of it.”
Traditional web scraping approaches were not viable. Selector-based code scripts are often brittle; a minor design change can break automation. Across hundreds of independently evolving sites, this would create ongoing maintenance overheads.
PX needed something more resilient: a system that could interpret websites more like a human would, finding relevant pages, understanding context, and determining whether required disclosures were present. The goal was to build an automated compliance audit capability that was reliable, extensible, and production-ready.
Solution
Delivering a resilient AI audit platform through Dialogue-Driven Delivery
From the beginning, we worked as one team. PX brought the domain knowledge (what “compliant” means in their world, and where the risk really sits), while our experienced consultants brought a delivery approach designed for speed without sacrificing engineering discipline.
“What I liked was that Equal Experts quickly created a lightweight proof point to show they understood the problem. That helped us sell it internally. Instead of waiting for instructions from us, the consultants worked alongside our teams to challenge our perceptions of the problem and help us improve our approach collaboratively,” said Mark van Heyningen, COS at PX.
We brought a lean team of just two consultants working full-time in an unfamiliar technical environment. A key enabler was Dialogue-Driven Delivery (D3). Instead of relying on long requirement documents, the team captured real conversations with PX stakeholders and used AI to turn those discussions into clear, testable specifications. That meant nuance wasn’t lost: edge cases, trade-offs and “what about…” questions were recorded and shaped into acceptance criteria that the team could implement and verify. With only two people in an unfamiliar environment, this approach reduced handoff overhead and kept decision-making tight.
On the technical side, what’s exciting here isn’t simply the use of AI. It’s the increase in reliability in a messy, constantly changing real-world environment.
Instead of writing fragile, page-specific scraping rules that break whenever a website changes, the team built an approach that handles variation by design. The system navigates sites more like a human would: it finds the right page, understands what it’s looking at, and extracts the relevant signals. That’s exactly what’s required when you’re auditing hundreds (and potentially thousands) of external websites that evolve independently.
That shift is important because it marks the difference between a proof of concept and something you can trust as an ongoing operational capability.
In practical terms, the system:
- Navigates lead forms and funnels autonomously
- Identifies privacy policy links regardless of placement
- Extracts “last updated” dates from differently structured documents
- Assesses whether required consent language is present
- Logs structured audit results
Equally importantly, this wasn’t treated as a one-off “smart script.” It was engineered as a repeatable product capability. The delivery approach translated stakeholder intent into clear, testable outcomes, and guardrails were put in place to protect quality and maintainability. Compliance checks were implemented as configuration rather than hardcoded logic, enabling PX to extend the system safely. Tight feedback loops, disciplined review practices, and workflow automation ensured the solution remained transparent and operable – not a black box.
Results
From manual review to scalable compliance capability
Within six weeks, PX had a working production system auditing real partner websites. Instead of spending time manually navigating individual funnels and hunting for disclosure text, the compliance team could shift their focus toward oversight and decision-making. Implementing compliance checks as configurable rules defined as data – not hardcoded logic – enables PX to add or modify audit criteria without code changes, maintain versioned compliance profiles and ensure consistency across audits.
Rules are transparent, adaptable, and easier to govern over time. PX has gained a scalable foundation for reducing operational risk as its partner ecosystem evolves. Just as importantly, the project demonstrated that AI can be applied responsibly even in a regulated environment, accelerating delivery while maintaining strong engineering discipline and quality standards.
Longer-Term Value
The most significant outcome was not just speed, but sustainable capability:
- A maintainable architecture
- Extensible rule configuration
- Production-grade infrastructure
- A proven model for AI-assisted delivery at speed
For technical stakeholders, the significance lies in the pattern this demonstrates: AI can be used pragmatically to reduce operational risk while preserving engineering discipline, clarity and long-term maintainability.
If you’re exploring how to modernise compliance processes, reduce operational risk, or apply AI in a disciplined, production-ready way, we’re always happy to share what we’ve learned from working alongside organisations like PX.