GPTs and Agents: Business Apps Reimagined
When Apple launched the App Store, businesses discovered something remarkable: you didn’t need to invent a whole operating system to serve your customers. You just needed an app that solved a specific problem — booking a cab, managing your calendar, or editing photos. Suddenly, developers could create small, focused pieces of software that lived inside a larger ecosystem. That shift unlocked entire industries.
We are standing at the same kind of inflection point with GPTs.
GPTs as the New Apps
Think of GPTs — OpenAI’s customizable models — as the equivalent of apps in the App Store. Instead of writing software in Python or Java, you “build” with business logic: documents, instructions, and processes. You tell the model what role it should play, what tone it should use, what steps it should follow. In other words, you configure it with the same kind of information your team already has in its playbooks and manuals.
This makes AI far more accessible than traditional software development. You don’t need an engineering department to get started. A consultant, a manager, or even a single subject-matter expert can now build a functional business app — in minutes.
“Building” Is Really Training
It’s useful to change our language here. We’re not coding; we’re training. When you design a GPT, you’re teaching it how to act in your business context. Just as you’d train a new hire with templates, examples, and workflows, you “train” an AI agent with prompts, sample documents, and clear boundaries.
That’s why the first wave of successful GPTs have looked a lot like assistants in very specific domains:
- Drafting and reviewing contracts
- Turning a marketing brief into a campaign outline
- Preparing product requirement documents (PRDs)
- Coaching junior team members through feedback and examples
Each one is essentially an app — but the “code” is business knowledge.
Why This Matters for Business Applications
Here’s the leap: if GPTs are apps, then the future of business software is less about downloading a platform and more about composing a team of AI agents. You’ll select GPTs like you once selected apps, slotting them into the workflows of finance, HR, product, and sales. Some will be off-the-shelf. Others you’ll “train” in-house, embedding your company’s data, language, and logic.
We’re still only scratching the surface. Today’s GPTs can handle document drafting, data summarization, and light process automation. Tomorrow’s will be embedded deeper into CRMs, ERPs, and collaboration tools, acting as colleagues rather than tools.
The ChatPRD Example
Take Claire Vo, the Chief Product Officer at LaunchDarkly. With a demanding day job, she still managed to build ChatPRD, an AI copilot for product managers.
She didn’t sit down to code a new software platform. She took the knowledge she already used daily — how to write PRDs, set goals, and evaluate metrics — and “trained” a GPT around it. Over a single weekend, she turned her expertise into an AI agent that now helps more than 30,000 product managers draft PRDs, brainstorm features, and even get coaching feedback.
ChatPRD is pulling in six-figure revenue, proving the point: you don’t need to build software the old way. You just need to package your business logic — the same instructions you’d give a new hire — into a GPT. That’s the new “App.”
And Claire did all this while keeping her full-time CPO role. Her story shows what’s possible when businesses realize they can create powerful, revenue-generating applications not by coding, but by teaching.
The Takeaway
The App Store democratized software distribution. GPTs are democratizing software creation. For businesses, that means “building” is no longer a technical exercise — it’s a strategic one. The question isn’t can we code this? It’s what knowledge and processes should we teach our AI agent so it becomes valuable to us and our customers?
The sooner businesses reframe “building” as “training,” the sooner they’ll unlock the next generation of business apps.
