Interest in AI is skyrocketing. Beyond the buzz, the real question remains: when – and how – should organizations use it?

At PLUS, we actively deploy AI as a tool, not a silver bullet. We have focused our efforts on custom-built GPT models to streamline operations, reducing time spent on manual tasks and freeing up valuable team bandwidth. But we’ve done it with a clear-eyed, test-and-learn approach – because AI isn’t magic (at least, not yet).

How to Approach AI: Measure Twice, Cut Once.

Our north star for any AI tool or solution is simple: does it help us work smarter?

Repetitive, time-consuming tasks are often ideal candidates for automation. We start by asking three simple questions:

• Is this a significant drain on time or team resources?
• Can an AI model handle the job at scale quickly and accurately?
• Can we prove our value in a test case before going further?

If the answer is yes, we start small, test rigorously and keep expectations realistic. We have learned a lot over the last year by implementing these safeguards and realized early on we aren’t chasing perfection – we’re chasing efficiency.

Case Study: AI-Driven Testimonial Management

One of our most impactful use cases? Transcribing and analyzing testimonials.

We developed a custom GPT to handle thousands of client-submitted stories, traditionally processed manually by multiple staffers on the team. Our model was trained to score relevance, tag by theme or key mentions and categorize based on the client goals.

Here are the results:

Key Takeaways

This AI solution wasn’t just faster, it was just as accurate as a manual scoring. The AI didn’t just save time, but it enhanced our ability to extract insights for our clients at scale by tagging and categorizing these testimonials. As a result, implementing this AI solution helped increase the bandwidth of multiple members of our team to focus on other client priorities.

Why This Worked

Careful work on the front end to program and construct a thoughtful AI platform ensured that it was looking for the right data. Instead of searching and searching for perfect video submissions — an impossible and unrealistic task, even for human brains — we asked our system to select submissions that checked as many important boxes as possible, a tailored approach to defining “accuracy.” We attached metrics to this method and saw the platform deliver measurable results from the outset. In the end, setting the right targets across all those videos made all the difference.

Our approach was simple and straightforward:

• Identify a repetitive task that was costing time across the team.
• Build a custom GPT trained on real sample data.
• Benchmark performance against manual output with realistic goals.
• Iterate to refine accuracy and relevance.

Because we approached this with intention and testing, we created a solution that not only works – but added value immediately to our projects.

Final Thoughts: Use AI to Evolve, Not Escape

AI can be an incredibly powerful tool for companies looking to grow and scale – but it won’t solve every problem. It’s not meant to. It’s meant to help you focus – to free up resources and time for the jobs that matter most.

At PLUS, we’ll continue to explore, test and adopt AI where it makes us better – not just faster. That’s how we stay agile, stay modern and stay ahead.

Back