Breaking the System to Make It Stronger: How AI Powers Our Pick Fest

By Bill Lucarell, Senior Software Engineer in Test, Lucas Systems

As a QA engineer at Lucas Systems, I believe that quality is paramount in every step of the software development life cycle. I focus on catching defects early and working with the developers to achieve quality so that warehouse managers are comfortable with their optimization suite. Some of the testing measures we take are:

• Developer-created unit tests
• Manual project-specific tests
• Automated system and integration tests
• Automated load and performance tests

Turning Testing into a Team Event

However, testing goes beyond the engineers’ manual and automated tests. Exploratory testing helps catch defects that planned testing does not. As a QA, I enjoy planning, executing, and communicating testing to team members. I envision the ideal software quality process as hands-on and learning-filled so that all understand how our software works and what constitutes good quality. Before the software reaches customer-lead user acceptance testing (UAT), we run it through multiple departments. This cross-team testing helps us find defects, identify feature ideas, and catch training issues ahead of time. We call this Pick Fest.

Before Pick Fest, I prepare a popup mock warehouse in the office “town hall” meeting area. Once it is set up, all department employees have a chance to test it, trying their hand to “break the system” in order to make it stronger and push it to the limits. Let’s look at some of the benefits a Pick Fest can give employees.

Uncovering Intangible Benefits Beyond Software Quality

I organize Pick Fests with a primary goal of catching hard-to-find bugs before they trickle to the customer. However, employees acquire more takeaways from Pick Fests.

First, the entire organization gets to see its product in action. We destroy knowledge silos and forge a collaborative work culture in the process.

Second, it is a training tool. New employees and non-engineers learn how the product works. They do so in a practical setting rather than solely reading product website pages and watching product demonstrations.

Third, it saves travel time and costs. Though warehouse visits are invaluable, Pick Fests provide an essential learning opportunity – one without paying for over 25 plane tickets and hotel rooms.

The Challenge: Setting up a Mock Warehouse Is Labor-Intensive

When I set up the “mock warehouse for the day”, I want “pickers” to work in a simulated environment mirroring a live warehouse. By running spontaneous workflows, employees:

• Stress-test automation — Uncovering bugs and inefficiencies before UAT
• Collaborate across roles — Engineers learn from customer success, and vice versa
• Onboard new employees — Giving them hands-on experience with real warehouse processes
• Build domain knowledge — Participants see how theory translates into practice

It’s a high-energy event that not only makes our software stronger but also strengthens our employees’ knowledge and morale.

Pickers could rely on a computer screen, but that removes physical constraints, human interaction, and real-time variability—key elements for realistic testing. Without these elements, we lose confidence that the system will run reliably in production. Edge and misuse cases are likely to be missed, lowering confidence of a quality solution.

By turning the town hall into a physical warehouse with location labels and “products”, such cases can be surfaced and tested. This gives us greater confidence in the solution before it reaches customers. Furthermore, employee pickers feel like actual warehouse workers, helping all departments understand a typical day in the warehouse. Many assume that setting up a physical environment requires significant time and effort, but with AI-driven tools, I make the process far easier than most would think.

The Solution: Supercharge the setup with AI

Without AI, setting up a mock physical warehouse requires a lot of busy work, such as:

• Writing or typing location labels
• Selecting product names and dimensions
• Manually bulk creating barcodes
• Storing observations from pickers during the Pick Fest
• Testers refining the structure of defects they insert

AI makes our setup more powerful by generating:

• Product and container barcodes at scale
• Location labels that match warehouse layouts and include checkstrings
• Synthetic but realistic inventory data
• Professional-looking instruction sheets to guide pickers
• A summary of observations from pickers as a template for UAT
• Well-structured defect reports after minimal user input

I used AI to reduce hours of manual setup to minutes. For example, I asked ChatGPT to generate a PDF with 50 Code 128 barcodes in a consecutive numeric range. I then specified the physical size, and it gave me a two-page PDF with all 50 barcodes so that they can be applied to containers. Because I automated with AI, the distractions were eliminated, and the focus was catching defects.

One important caveat is to use secure AI tools with AES-256 or similar encryption. These tools give us peace of mind knowing that their data is protected.

Bonus #1: AES-256 Encryption protects company data

When preparing a Pick Fest with AI, I always avoid the free tools and use AI with AES-256 encryption. ChatGPT Business and Microsoft Copilot both offer this encryption.

Imagine company data in an “encrypted fortress” – so secure that an army using brute force would take longer than the age of the universe to attack. This encryption gives us confidence that internal data is not being used to train external data models.

Bonus #2: Incorporating gamification into Pick Fests

I am passionate about making work fun while achieving quality. According to a Lucas Systems study, 84% of warehouse workers are more inclined to stay at their workplace if their workplace incentivizes work with gamification.

At a recent Pick Fest, we did that. Whoever found the most defects won a gift card, which further drove the goal of not letting any defect escape to the customer. Everyone was more motivated to find issues scripted tests miss. As a result, employees learned about the product, collaborated well, and had fun!

Real-World Impact: Building Confidence Before Go-Live

By the time our software goes to the customer, it has already endured stress tests in our mock warehouse during Pick Fest. When I travel to a customer, user acceptance testing is all but a non-event and workers feel comfortable using the system. At one warehouse, productivity jumped 30 percent four weeks after our software replaced paper picking.

For us, Pick Fest doubles as a training module — turning abstract features into hands-on experience and making us stronger advocates for our customers.

Conclusion: Breaking to Build Better

At Lucas Systems, we believe the “try to break things — together” philosophy is essential to quality software and team building. People, AI, and technology combine to form a realistic warehouse environment that mimics Lucas’s customers.

As an example, we recently upgraded a customer to the latest version of warehouse automation software. I organized a Pick Fest where the mock warehouse was built with AI, and I involved two new employees learning the software while contributing their findings. As a result, Lucas implemented the upgrade in record time, and two employees gained invaluable domain knowledge – a win-win!

Without these findings, customers might question the quality of their solution. Instead, customers feel secure knowing that everyone understands the solution and is committed to defect-free software.

Bill Lucarell is a Senior Software Development Engineer in Test at Lucas Systems, where he leads testing and automation processes. He brings over 15 years of software industry experience to Lucas and has worked as a developer and a test automation engineer for various B2B products. After learning Agile in 2010, he became a champion for Agile processes and leverages those processes to produce high quality software and build team camaraderie.