How Real Companies Are Using AI to Build Products, Not Just Write Emails
Most of the AI conversation in small business circles is still stuck at the surface level. "Use ChatGPT to write your captions." "Ask Claude to summarize your meeting notes." That's fine. It's a start. But the companies that are actually pulling ahead right now are using AI to do something bigger: build entire product lines faster than they could have staffed a team to do it.
A recent episode of Lenny's Newsletter broke down exactly how Gusto did this, and there's a lesson in it that applies directly to anyone running a small business, a consulting practice, or a service-based operation.
What Gusto Actually Did (And Why It Matters to You)
Gusto, the payroll and HR platform, used Claude Code to build a new product line. Not to polish an existing feature. Not to speed up a dev sprint. To create something net new that didn't exist in their roadmap before AI made it feasible.
They gave Claude Code the context it needed, the business logic, the constraints, the outcome they were after, and the tool did a significant portion of the build. Their engineers weren't replaced. They were repurposed. Instead of writing boilerplate and debugging routine code, they were reviewing, guiding, and making judgment calls.
That's the shift worth paying attention to. AI didn't replace the decision-making. It eliminated the grunt work that was slowing the decision-making down.
For a small business owner, the parallel is direct. You probably have a product idea, a service offering, or a workflow you've been sitting on because you don't have the bandwidth to build it. AI doesn't give you a developer. But it does give you the ability to move faster with whatever capacity you already have.
GLM-5.2: Another Model Worth Knowing
The same source also covered a review of GLM-5.2, a Chinese-developed large language model from Zhipu AI. It's been gaining traction as a capable, cost-efficient alternative in the AI model landscape.
Here's why this matters even if you never plan to use GLM-5.2 directly. The pace at which capable models are releasing right now is compressing the timeline on everything. Every few months, a new model arrives that can do something the previous generation couldn't, or can do the same thing for a fraction of the cost.
That means if you built an assumption that "AI tools are too expensive for my business," that assumption is already stale. And if you've been putting off experimenting because you weren't sure which tool to commit to, the answer is: don't commit to one. Build workflows that are model-agnostic where you can.
The businesses winning with AI aren't loyal to a single tool. They're loyal to the outcome and they use whatever gets them there efficiently.
The Actual Takeaway for Your Business
The Gusto story isn't about a big company doing a big thing. It's about a principle: AI lowers the minimum viable team size for building something new.
Six months ago, launching a new product line or service tier required headcount you probably didn't have. A developer, maybe a project manager, definitely a few months of runway. Now, with tools like Claude Code or even standard Claude or ChatGPT with the right prompting, you can prototype something, test the logic, draft the delivery mechanism, and validate it with a small group before you've spent anything meaningful.
That's not a small shift. That's a structural change in what's possible for operators running lean.
If you're a consultant, this means you can package IP you've never had time to systemize. If you're a coach, it means you can build a digital resource you've been sketching in notebooks for two years. If you run a service business, it means you can finally document and automate the onboarding process you've been meaning to build since you hired your second employee.
Where to Start This Week
Don't try to replicate what Gusto did. That's not the point.
The point is to identify the one thing in your business that you haven't built yet because it felt too time-consuming to set up. One deliverable. One system. One offer that exists in your head but nowhere else.
Open Claude or ChatGPT. Describe what you want to build, in plain language, like you're explaining it to a smart contractor. Include the purpose, the audience, and the constraints. Then ask it to give you a first draft, a structure, or a step-by-step build plan.
You're not outsourcing your judgment. You're outsourcing the blank page problem.
The companies moving fastest right now aren't the ones with the biggest teams or the most sophisticated AI stacks. They're the ones who stopped waiting for the perfect setup and started building with what's already available.
You already have access to the same tools Gusto used. The question is what you're going to build with them.
Pick one thing. Start today.