It’s quite clear that agentic coding has completely taken over the software development world. Writing code will never be the same. Shoot, it won’t be long before we aren’t writing any code at all because agents can write it better and faster than we humans can. That may already be true today. 

But there is more to software development than merely writing code, and those areas—source control, documentation, CI/CD, project management—are ripe for some serious disruption from AI as well. Those areas may well be hit harder than coding itself. 

I would imagine that if you were in the business of analyzing data and providing dashboard-level insights into that data, then you would be very worried indeed about what AI is going to do to your value proposition. Much of the SaaS industry is in the business of analyzing existing data, and that is exactly what AI agents can do well. When a simple question can get straight to the heart of what a pricey dashboard provides, then companies have to question the value of paying for that kind of service.  

Tools like LinearB, Jellyfish, and Swarmia provide deep and interesting insights into what is going on inside your repository, but if you can say to Claude Code, “What are the DORA metrics for this repository?”, well, then those businesses are definitely ripe for disruption, no? 

Pivoting to AI

Those tools are already reacting by pivoting hard and leaning into the AI revolution. They are doing things like focusing on measuring AI processes instead of providing team insights. These tools are now pitching that they monitor not your development team but your AI development process, which is the kind of thing they have to do when the ground under their feet is shifting. The disruption is real, and they have to change or die. 

Dashboards over existing data need to make a rapid change. But tools that produce underlying data need to change as well. Instead of producing dashboards for human consumption, these tools are turning hard towards providing Model Context Protocol (MCP) implementations that AI agents can consume.  

One meta-coding area where I have found AI provides real value is in log examination. When a problem occurs, the first question that usually gets asked is, “Where is the log of that happening?” Back in the before times, you’d have to pore over the log, line by line, searching for exactly what happened for clues into the source of the problem. But now? Give the log, however large, to an AI agent, and those answers appear in a matter of minutes. 

Producing the log becomes the real value—displaying dashboards over that data becomes less important. A tool like Datadog owns the ingestion pipeline and the time-series production, and it creates valuable data, so its pivot is easier. Datadog need only create a tool that talks to an AI agent instead of a human. Their beachhead is solid. The real value of logs lies in an agent’s ability to peer into them in real time and take action based on what it sees. It won’t be long until, whenever a problem occurs, an MCP server will notify an AI agent and the agent will analyze the problem, fix it, and deploy the fix, all without human intervention.  

Producing and owning the data beats being able to interpret the data. Tools that produce the data can lean into the AI revolution. Tools that merely read and display data from a different source—say, an existing repository—will have a much harder time surviving alongside AI agents. 

The soul of a new user

Any provider of a software tool that is part of a development or operations workflow should be working very hard to provide an MCP or a CLI for an AI agent to use, because that is the future. A CI/CD system needs to be able to respond to events without a human being involved at all. Such tools become the data source and will have an entirely different front end. Instead of humans looking at dashboards, it will be AI agents making MCP queries into the tool. 

This is where the disruption is really happening. One might even say your customer is no longer a software development manager but an AI agent’s MCP server. How long will it be before we have AI tools making purchasing decisions after running thousands of simulations against a set of potential new tools? Previously, software tool companies put a lot of energy into slick-looking UIs, web pages with solid copy, and all kinds of bells and whistles meant for human consumption. 

But does any of that matter if you are actually selling to an AI agent? Does your MCP server actually return data that another MCP server can consume and use? 

Everything that SaaS companies have learned to do to be successful is now being turned on its head. AI agents don’t care one whit about cool-looking websites and clever marketing copy. Selling to a machine that doesn’t care about your pitch, your carefully crafted brand, or your clever logo is a game that no one has ever played before.