AI is everywhere right now, and it’s easy to see why. It’s leveling the playing field for small businesses, enabling them to compete at scale and improve how they serve their customers. AI also helps make sense of data to quickly identify trends and opportunities, and it handles repetitive tasks so people can focus on the bigger stuff. The potential is huge.
But here’s the thing: As exciting as AI is, it’s not magic, and it’s not a replacement for people. It’s only as good as the thought and care humans put into it. If we get it right, it has the potential to solve real problems for real people by helping them augment their work and their ability to build new things.
At Gusto, we support over 400,000 small and mid-sized businesses with HR, payroll, and benefits. The businesses we serve have big dreams but limited resources, and many are already using AI to scale their work. We’ve taken a deliberate approach to integrating AI with our product and operations, using human-centered design to inform how and where it appears inside Gusto.
For companies building their own AI-powered solutions, it’s important to balance excitement with responsibility. Here are three foundational principles we’ve developed to get the most out of AI while keeping humans at the heart of the process.
1. TAP INTO HUMAN JUDGMENT AND EXPERTISE TO PREVENT ERRORS
One core concept for how we design for AI is “human in the loop.” This means having humans review, adjust, and make final decisions on what AI has provided them. Without this oversight, AI can make mistakes and overlook nuances.
We saw this firsthand when we built Gus, our AI assistant. It can be overwhelming for business owners to navigate tasks like filing taxes and hiring employees. We designed Gus to help businesses get fast answers in plain, simple language for their most common payroll, benefits, and HR questions.
As we continue to iterate on Gus, we’ve used “human in the loop” to bring in human experts—including our legal, compliance, and HR teams—to verify Gus shares the right information and to fine-tune its results. We also tapped our customer support teams to ensure the most common customer questions had solid, reliable responses.
Involving humans as a core part of the design process helps reduce the risk of errors while building a better experience for small business owners—the real people who actually use Gus.
2. EMPOWER MORE PEOPLE BY MERGING THEIR EXPERTISE WITH AI
Applying “human in the loop” isn’t just important for preventing errors. It’s also a catalyst empowering humans to build new things.
People across roles often have great ideas, but traditionally, if they didn’t have the technical skills to bring them to life, their ideas could stall out. With the right tooling and infrastructure in place, AI can be an enabler. Humans bring subject matter expertise, customer centricity, and judgment to define what should be built, and AI democratizes some of the technical execution required to build what they’ve envisioned.
For example, we recently launched a proprietary tooling that enables non-technical customer support employees to create simple AI “recipes.” These quickly pull from our internal knowledge base and generate first drafts of answers to common customer questions. Other employees can review, edit, and repurpose these recipes.
Today, some of these tools receive 2,000 to 3,000 daily interactions from our customer support teams. On average, they’ve reduced the time spent answering similar customer questions by 33%. Empowering more creativity within a company—with the right guardrails in place—can unlock incredible customer experiences.
3. GIVE HUMANS THE FINAL SAY IN THE DECISION-MAKING PROCESS
The last piece of “human in the loop” involves making people the final decision-makers. This was critical for us in building our AI assistant.
Gus can provide businesses with personalized insights and execute manual tasks, like approving time off requests or drafting job postings. These are useful and time-saving, but small business owners need to be in control.
If they ask Gus to generate a first draft of a job posting, Gus will allow them to edit it and then give final approval before posting. Or if they ask Gus to approve a time off request, or change someone’s manager, Gus will confirm the action before taking it so they can provide final approval. The technology isn’t a replacement for humans—it enables business owners to make the final decisions for their unique needs.
KEEPING HUMANS AT THE CENTER OF AI EVOLUTION
AI is ultimately about augmenting, not replacing people. As it evolves, it will be more important than ever for companies to be intentional about the role that humans play, and to ensure their experiences put humans—and their real needs—at the center.