Who builds your software now, and who earns from it

Two forces turned up in the same day's news, and they point the same way. Artificial intelligence is changing both who can build software and how much money the companies selling it now make.

For a business choosing tools, that is not abstract. It changes the build-or-buy question, it changes which suppliers are safe to depend on, and it changes how you plan headcount against automation. Below are the items from the day that carry the most weight for those decisions, in order of how much they should affect a buyer.

The people building software no longer all write code

For most of the history of business software, building an application meant employing people who could write code. If you could not, you bought something off the shelf and lived with its shape. That line is moving. Zapier reported that 34% of people shipping software using AI tools now have no formal programming background [1]. The piece notes that the technology has moved past the point where only trained developers benefited from it: "AI was impressive enough when skilled developers could use ChatGPT to troubleshoot code" [1], and the point of the article is that the group who can now ship working software is much wider.

What should a buyer take from this. First, the build-or-buy line has shifted, but not as far as the headline suggests. Being able to assemble a small internal tool is not the same as running one safely for years: someone still has to own the data, the access rules, the backups, and the day the thing breaks. The right question is not "can we build this" but "who maintains it in eighteen months, and what happens if that person leaves." That is the same discipline you should apply to any tool you pay for, which is the subject of choosing software worth using. Second, the fact that a task *can* be automated or self-built is not a reason to do it. The test is whether the work is repetitive, well-defined, and low-risk when it goes wrong, which is covered in how to tell whether a task should be automated.

An AI supplier is now among the largest earners in software

When you choose a tool, you are also choosing a supplier, and it helps to know how big and how fast-growing that supplier is. SaaStr made the scale concrete: it argued that by year-end one AI company will out-earn every public software company except one very large incumbent, noting that "Anthropic exited 2025 at roughly $9 billion in run-rate revenue" [2].

The number matters less than what it implies for a buyer. A supplier growing at that rate is not going to disappear, which reduces one kind of risk. But rapid concentration brings a different risk: pricing power, and the temptation to build your workflow so tightly around one provider that leaving becomes expensive. The practical response is not to avoid large suppliers. It is to keep asking a boring question about every tool: if this vendor doubled its price or changed its terms, what would it cost us to move. That question is easier to answer when you have decided in advance what AI is allowed to touch in your business and what it is not, which is the ground covered in what AI should and should not do in your business.

Layoffs, and the question sitting underneath them

The same day brought news that a large technology firm cut staff. TechCrunch reported that "Microsoft cut around 4,800 roles, or 2.1% of its global workforce, on Monday — the latest in a series of layoffs that's stoking fears of AI replacing jobs" [3], and that the reductions would fall hardest on its Xbox and commercial sales groups [3].

Read this carefully rather than as a verdict. The report states that the cuts are stoking fears about AI and jobs; it does not state that AI caused these particular cuts, and a briefing should not claim more than the source does. What is useful for a smaller business is the underlying decision that every layoff and every hire represents: when is the right answer a person, and when is it a process. Cutting people because a tool exists, or hiring people because a process is painful, are both mistakes made in the same place. The reasoning that keeps them apart is set out in when to hire and when to automate. The short version is that you automate stable, repeatable work and you hire for judgement, relationships, and the parts of the job that change every week.

Tools that can join the meeting you already run

A quieter announcement carried a lesson that outlasts any single feature. Google Workspace said that "You can now join video conferences on Google Meet hardware via SIP through a Pexip interop gateway" [4], and that this lets people "join meetings hosted on any SIP-compatible platform directly from their Meet rooms" [4].

The detail to notice is the standard. SIP is an open protocol, and the news here is one platform's hardware reaching meetings run on other platforms through it. For a buyer, interoperability through shared standards is worth more than any single integration, because it is what stops a tool from becoming an island. When your calendar, your meeting room, and your customer records cannot reach each other, your team pays for it every day in copied information and repeated questions — the cost described in why your tools do not talk to each other. When you weigh a new tool, ask what it speaks, not just what it does. A product that talks to what you already run is worth more than a slightly better product that only talks to itself, which is one reason 360REV keeps a customer's conversations, quotes, and records on a single timeline rather than in separate apps.

Software that has to grow with you

The last item is about fit over time. Xero published a note on building its product to scale with growing businesses, and it opened on the moment most buyers underestimate: "There's a moment in the life of a growing business that rarely makes it into the headline" [5] — the point at which things are going well and the complexity of running the business starts to outpace the tools that got it there.

That moment is worth planning for before you reach it. A tool that fits ten customers may not fit a thousand, and the switch is always more expensive when it is forced rather than chosen. This is the same pattern as the day a spreadsheet stops being enough: nothing breaks loudly, the work just gets slower and more error-prone until someone admits the tool has been outgrown, a transition examined in when a spreadsheet stops being enough. The lesson from the vendor's own framing is to choose tools that have a credible path from your size now to your size later, so that success does not force a migration at the worst possible time.

What the day adds up to

The thread across all five items is the same one that ran through yesterday's briefing: AI is lowering the cost of building software while raising the stakes of the suppliers you depend on. For a business choosing tools, none of this changes the fundamentals. Decide what work is worth automating, know how hard it would be to leave any supplier you lean on, buy tools that talk to each other through open standards, and choose for the size you are growing into. The technology moves quickly. The questions a careful buyer should ask do not.