It’s fairly clear that agentic coding has fully taken over the software program improvement world. Writing code won’t ever be the identical. Shoot, it received’t be lengthy earlier than we aren’t writing any code in any respect as a result of brokers can write it higher and quicker than we people can. Which will already be true in the present day.
However there may be extra to software program improvement than merely writing code, and people areas—supply management, documentation, CI/CD, venture administration—are ripe for some severe disruption from AI as properly. These areas might be hit tougher than coding itself.
I might think about that when you have been within the enterprise of analyzing knowledge and offering dashboard-level insights into that knowledge, then you definitely could be very apprehensive certainly about what AI goes to do to your worth proposition. A lot of the SaaS business is within the enterprise of analyzing current knowledge, and that’s precisely what AI brokers can do properly. When a easy query can get straight to the guts of what an expensive dashboard gives, then firms must query the worth of paying for that sort of service.
Instruments like LinearB, Jellyfish, and Swarmia present deep and attention-grabbing insights into what’s going on inside your repository, however when you can say to Claude Code, “What are the DORA metrics for this repository?”, properly, then these companies are positively ripe for disruption, no?
Pivoting to AI
These instruments are already reacting by pivoting arduous and leaning into the AI revolution. They’re doing issues like specializing in measuring AI processes as a substitute of offering staff insights. These instruments at the moment are pitching that they monitor not your improvement staff however your AI improvement course of, which is the sort of factor they must do when the bottom below their ft is shifting. The disruption is actual, and so they have to vary or die.
Dashboards over current knowledge must make a fast change. However instruments that produce underlying knowledge want to vary as properly. As an alternative of manufacturing dashboards for human consumption, these instruments are turning arduous in direction of offering Mannequin Context Protocol (MCP) implementations that AI brokers can eat.
One meta-coding space the place I’ve discovered AI gives actual worth is in log examination. When an issue happens, the primary query that often will get requested is, “The place is the log of that occuring?” Again within the earlier than instances, you’d must pore over the log, line by line, looking for precisely what occurred for clues into the supply of the issue. However now? Give the log, nevertheless massive, to an AI agent, and people solutions seem in a matter of minutes.
Producing the log turns into the actual worth—displaying dashboards over that knowledge turns into much less vital. A software like Datadog owns the ingestion pipeline and the time-series manufacturing, and it creates priceless knowledge, so its pivot is simpler. Datadog want solely create a software that talks to an AI agent as a substitute of a human. Their beachhead is strong. The actual worth of logs lies in an agent’s means to look into them in actual time and take motion primarily based on what it sees. It received’t be lengthy till, each time an issue happens, an MCP server will notify an AI agent and the agent will analyze the issue, repair it, and deploy the repair, all with out human intervention.
Producing and proudly owning the info beats with the ability to interpret the info. Instruments that produce the info can lean into the AI revolution. Instruments that merely learn and show knowledge from a distinct supply—say, an current repository—may have a a lot tougher time surviving alongside AI brokers.
The soul of a brand new person
Any supplier of a software program software that’s a part of a improvement or operations workflow ought to be working very arduous to supply an MCP or a CLI for an AI agent to make use of, as a result of that is the longer term. A CI/CD system wants to have the ability to reply to occasions with out a human being concerned in any respect. Such instruments develop into the info supply and may have a wholly totally different entrance finish. As an alternative of people taking a look at dashboards, it is going to be AI brokers making MCP queries into the software.
That is the place the disruption is admittedly taking place. One would possibly even say your buyer is now not a software program improvement supervisor however an AI agent’s MCP server. How lengthy will or not it’s earlier than we’ve AI instruments making buying choices after working hundreds of simulations in opposition to a set of potential new instruments? Beforehand, software program software firms put quite a lot of vitality into slick-looking UIs, net pages with strong copy, and all types of bells and whistles meant for human consumption.
However does any of that matter in case you are really promoting to an AI agent? Does your MCP server really return knowledge that one other MCP server can eat and use?
Every little thing that SaaS firms have realized to do to achieve success is now being turned on its head. AI brokers don’t care one whit about cool-looking web sites and intelligent advertising copy. Promoting to a machine that doesn’t care about your pitch, your fastidiously crafted model, or your intelligent brand is a sport that nobody has ever performed earlier than.
