Now Everyone's a Software Builder & Companies Need to Reprice

In a time and land not so far away, building a app, video game, downloadable song, website, or integration between two pieces of software required a very specific kind of person. A programmer, someone who had spent years learning to code. Someone who spoke the languages of the machine: Python, JavaScript, PostgreSQL, REST API's etc. The rest of us had ideas, and those ideas stayed ideas unless we pad a programmer to build them into reality for us.

That era is over.

Agentic AI is making it easy for anyone to build apps, games, songs, videos, websites, and integrations. Not "easy" in the marketing-brochure sense. Easy in the literal, everyday, I-did-it-this-week sense. And the structural foundation that has made all this possible is API calls.

The Key That Unlocked Everything: API Calls

An API — an Application Programming Interface — is a doorway between two pieces of software. It's a structured, documented way for one program to ask another program to do something: read this data, create that record, send this message, update that status. Every time a program makes one of those requests, that's an API call.

Now AI agents can make API calls. An agent doesn't just generate text — it can act through an API call. It can reach through the doorway of a platform and do real work inside it, change setting, create, move data in and out etc. When you chat with an agent and say "create a dashboard that gives me a summary of all the pulses, emails, and tickets that I need to follow up on, with a brief summary of why" the agent translates your sentence into the right API calls, sends them to the right platforms that you have connected or given your agent access to and executes what you want.

The gap between "what I want" and "the technical steps to make it happen" has collapsed to zero. You no longer need to know the syntax, the tech stack, or how programming works in general for the most part. You just need to know what you want to build or do.

Three Doors, Three Examples

Slack. Slack's API lets an agent read channels, post messages, create canvases, react to threads, and send notifications. So an agent can watch for something important happening in one platform and alert your team in Slack the moment it happens. "When you get an alert that pulse is ready for QA testing from Monday.com, alert me in this thread." That's an integration that allows for a fast work through through a single alert.

Monday.com. Monday has a full API, which means an agent connected to Monday can read every pulse on your board, create new items, update statuses, attach files, move things between boards, and write results back. In my own work, I ask my agent to connect to Monday, analyze all the pulses, and give me a detailed, straight-to-the-point report on any topic or trend. I also had it compare upcoming deploys against the knowledge base articles I manage and tell me which ones need updating based on which pulses are in the deployment. What would have required a developer writing scripts — or me building a spreadsheet and manually updating it (which I actually had already done before agentic AI came out), taking hours of work — happens in one conversation.

Lovable. Lovable takes it a step further — it's a full website and app builder where you literally chat. You describe what you want — "build a website that is by invitation only that I can share with my coworkers that records all the feature requests that my company receives through "X" means, includes a sliding scale that my teammates can move back and forth as they hear about feature requests through meetings or webinars, and add columns that show an estimated development time to build it based on my company's programming architecture, QA time, whether a KB articles needs to be updated or created, if support and sales needs a formal meeting to learn the learn feature of just be given a write up etc and make the page auto sort so the most requested feature requests are always at the top" — and it builds it, styles it based on the guidelines you give it, and deploys it with one click of a button. This is how people can build whole apps and websites (like DuckBuddy.net) without writing a single line of code. All you need to be is creative, you don't have to be a developer.

Notice what all three have in common: a doorway (the API), an actor (the agent), and a person who just talks. That's the new formula for building software.

The Technical Name for the Magic

When an AI agent uses a connector to interact with an external platform or service to perform a task, the action is called tool calling (or function calling). The agent doesn't just chat — it invokes a tool: "post to Slack," "create a Monday pulse," "deploy a page." Meanwhile, the actual bridge or integration layer — the thing that connects the agent to the platform — is referred to as an agent connector or AI connector.

Tool calling is the verb. The connector is the doorway. Once you know those two words, you have a basic understanding for how this magic is working behind the scenes.

The Race Resulting in a Cultural Shift

We've all watched the race happen. Companies are sprinting to put AI into their products — whether it genuinely helps or not. Smart TVs have AI. Phone apps have AI. Platforms like Monday.com have their own agent baked in. Website builders like Lovable are built around AI. Reddit has AI answers and AI powered search engines. There's even an AI bird feeder. I'm not joking, and that's exactly the point: AI is being crammed into our lives everywhere, useful or not.

But here's the deeper truth underneath the hype cycle: all of that cramming has shifted our culture. We now expect AI everywhere. And not just on gadgets — in our work software. The people who are on board with using it now expect every software they use at work to have its own agent, or to be able to connect to their agent, so they can talk to it and have it perform tasks in that platform.

Ask people what they expect AI to do at work and the answer is ambitious: analyze large amounts of data, find trends in sales, dig through customer profiles, surface shopping trends — so their company can be more successful at reaching customers. Whether every product lives up to that expectation is another question. But the expectation itself is now the standard.

There's also a perception problem — or maybe a perception reality: people believe AI can do basically anything. And honestly, they can be forgiven for it. Scroll social media and the AI-generated content is everywhere, and it looks real — videos, images, music. Videos, papers, articles (including the one you're reading, though I'll be honest: I give very lengthy prompts to make sure these turn out the way I want, and I still edit them), books — all being created with AI. People can crank out content at a volume that used to require a studio or a publishing house.

And underneath the content explosion is a code explosion. Developers and QA testers are generating tons of code, code tests, and performing those tests at triple or quadruple the speed of someone programming and testing manually. The productivity curve hasn't just bent — it's a different curve entirely.

No One Has to Wait for Feature Requests Anymore

With the widespread capabilites of AI, People no longer have to wait for a feature request of theirs to be implemented by a company. They can ask their agent to build it for them — as an external tool, a dashboard, an alert system, whatever shape the need takes.

Think about what that means. For decades, the deal was: you buy software, it does what it does, and if you want it to do something else, you file a request and wait. Maybe it gets built next quarter. Maybe next year. Maybe never.

Now? If Monday doesn't have the report I want, my agent builds the report by reading Monday's data through the API. If Slack doesn't have the alert I want, my agent creates the alert flow. People are having their agents build tools and manipulate settings in the software they're using in order to get exactly what they want — and they're having their agents build integrations between the different softwares they use, chaining each platform's features together with added capability. Figma to Jira. A calendar app to an HVAC system. If both sides have a doorway, an agent can walk through both.

That is a fundamental change in how people use software. We went from consumers of features to commissioners of our own.

Add-On Sales are Down, API Calls are Up

As this trend grows, companies may see a downtrend in "add-ons" to their product — those extra paid features and packages that customers used to buy when the basic or core product alone didn't do enough. Why buy the premium reporting add-on when your agent can build you a better report from the API?

But at the same time, they'll see an uptick in the API calls their customers are using. Masses of them. Every agent-built integration, every automated dashboard, every cross-platform pipeline — it's all riding on API calls.

For the end user, this is mostly great. You get exactly what you want, tailored to your workflow, without waiting for a product team to prioritize you. But it forces a real change on the software companies: they now have to change how they charge for their product — both to keep up with the demand and usage patterns of their customers, and simply to stay profitable.

The mechanics are unforgiving. API usage isn't free for the provider. Every call hits their servers, their infrastructure, their bandwidth. Many companies have charged extra for API access, and I can see that cost increasing (if it hasn't already) as AI-driven call volume explodes — the same way token usage on AI models drives cost. The more agents the world runs, the more calls everything makes, and someone has to carry that load.

We've already watched this movie. People started off using ChatGPT for free, and the demand and usage was so high that OpenAI decided to — had to — start charging for it. We see the same trend across AI models: recently DeepSeek raised their prices substantially on OpenCode after being one of the cheapest options on the platform. That's not greed, or at least it's not only greed. Once a service gets heavy usage, companies need to charge more to handle the load on their servers — and in principle, the price of things in demand goes up. That's Economics 101 wearing a new hat.

Value Is Moving and so it the Charging of Value

For years, the value lived in add-ons — premium features, packages, upgrades. Customers found value in what the product could do more of, and they paid for it.

Now the value is migrating. Customers are finding value in access — in the API doorway being wide open, the connectors being rich, the platform being friendly to their agents. The value moved from "what your software does" to "what my agent can do with your software."

So pricing will follow. Companies will start charging for their value according to where customers are actually finding it: in API usage, in connector access, in agent-friendly infrastructure. Some will do it gracefully, with usage tiers that scale with what you build. Some will do it clumsily, with rate limits that feel like a tollbooth on a highway they just discovered everyone is driving on. But all of them will do it, because that's where the product is now being consumed.

Bottom Line

The age of the builder isn't coming. It's here. A person with an idea and a chat window can now build the app, the game, the website, the integration — the thing that used to live and die as a feature request. The doorway was always the API; what's new is that agents have made the doorway usable by everyone, and tool calling has turned conversation into construction.

But every revolution rearranges who pays for what. As we all stop buying add-ons and start consuming API calls — as our agents quietly make thousands of requests on our behalf every day — the software industry's pricing model will bend around the new source of value. Free things will get expensive, cheap things will get pricier, and companies will charge where the value has moved. We'll grumble about it, and then we'll pay it, because we won't want to go back to waiting for features any more than we'd go back to waiting for letters.

The real skill, then, isn't learning to build. The agents handle that. The real skill is knowing what's worth building — and watching, with clear eyes, as the bill for our new superpower starts arriving in the form of API usage on our invoices.

We asked for the power to build anything. We're getting it. And like everything worth having, it's going to show up on a pricing sheet.

Anyone can be a builder now. Anyone can create, "program", or build an integration and in order to stay in business, companies will need to rethink how they charge customer, some already have...