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 gained’t be lengthy earlier than we aren’t writing any code in any respect as a result of brokers can write it higher and sooner than we people can. Which will already be true in the present day.
However there’s extra to software program improvement than merely writing code, and people areas—supply management, documentation, CI/CD, mission administration—are ripe for some critical disruption from AI as effectively. These areas could be hit more durable than coding itself.
I’d think about that for those who had been within the enterprise of analyzing knowledge and offering dashboard-level insights into that knowledge, then you definitely can 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 effectively. When a easy query can get straight to the guts of what an expensive dashboard supplies, then firms should query the worth of paying for that type of service.
Instruments like LinearB, Jellyfish, and Swarmia present deep and fascinating insights into what’s going on inside your repository, however for those who can say to Claude Code, “What are the DORA metrics for this repository?”, effectively, then these companies are undoubtedly ripe for disruption, no?
Pivoting to AI
These instruments are already reacting by pivoting laborious and leaning into the AI revolution. They’re doing issues like specializing in measuring AI processes as an alternative of offering group insights. These instruments at the moment are pitching that they monitor not your improvement group however your AI improvement course of, which is the type of factor they should do when the bottom beneath their toes is shifting. The disruption is actual, they usually have to alter or die.
Dashboards over current knowledge have to make a fast change. However instruments that produce underlying knowledge want to alter as effectively. As a substitute of manufacturing dashboards for human consumption, these instruments are turning laborious in the direction of offering Mannequin Context Protocol (MCP) implementations that AI brokers can devour.
One meta-coding space the place I’ve discovered AI supplies actual worth is in log examination. When an issue happens, the primary query that normally will get requested is, “The place is the log of that taking place?” Again within the earlier than occasions, you’d should 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, nonetheless massive, to an AI agent, and people solutions seem in a matter of minutes.
Producing the log turns into the true worth—displaying dashboards over that knowledge turns into much less vital. A device like Datadog owns the ingestion pipeline and the time-series manufacturing, and it creates invaluable knowledge, so its pivot is less complicated. Datadog want solely create a device that talks to an AI agent as an alternative of a human. Their beachhead is stable. The true worth of logs lies in an agent’s means to see into them in actual time and take motion primarily based on what it sees. It gained’t be lengthy till, every 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 information beats with the ability to interpret the information. Instruments that produce the information can lean into the AI revolution. Instruments that merely learn and show knowledge from a unique supply—say, an current repository—can have a a lot more durable time surviving alongside AI brokers.
The soul of a brand new consumer
Any supplier of a software program device that’s a part of a improvement or operations workflow needs to be working very laborious to offer an MCP or a CLI for an AI agent to make use of, as a result of that is the long run. 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 information supply and can have a completely totally different entrance finish. As a substitute of people taking a look at dashboards, it is going to be AI brokers making MCP queries into the device.
That is the place the disruption is basically occurring. One may even say your buyer is now not a software program improvement supervisor however an AI agent’s MCP server. How lengthy will it’s earlier than we have now AI instruments making buying selections after operating 1000’s of simulations in opposition to a set of potential new instruments? Beforehand, software program device firms put a variety of power into slick-looking UIs, net pages with stable copy, and every kind of bells and whistles meant for human consumption.
However does any of that matter if you’re really promoting to an AI agent? Does your MCP server really return knowledge that one other MCP server can devour and use?
Every part that SaaS firms have discovered 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 rigorously crafted model, or your intelligent emblem is a sport that nobody has ever performed earlier than.