AI is real, and it is already inside your ICP’s stack

I read a lot of market data. Most of it is noise. Every quarter someone declares a new era, and most of the time the numbers underneath the headline are thin.

This quarter is different. The signal is loud, and if you sell into the enterprise like I do, it is pointed straight at your buyer.

Here is what I mean.

The money is voting, and it is voting for agents

US software M&A ran close to $248b in a single quarter. That number is inflated by a handful of mega deals, so I do not treat it as the whole story. What I do treat as the story is what got bought and why.

Salesforce paid $3.6b for an AI customer service agent. A private equity firm took an agentic ServiceOps platform at 10x EBITDA. A coding assistant sold for $60b at sixty times revenue. Strip out the eye-popping multiples and look at the category underneath each deal. Customer service. IT service operations. Developer workflows. Finance data.

Those are not fringe categories. Those are the exact systems a CIO, a CFO, and a CISO are responsible for right now.

Here are the ten largest disclosed deals of the quarter. Read the target descriptions, not just the dollar amounts. Almost every one of them is AI, data, or automation aimed at an enterprise workflow.

Nine of the ten are either an AI product, the data plumbing that feeds one, or the compute underneath it. That is not a coincidence. That is where the enterprise dollar is going.

This is not disruption in the abstract

For two years the AI conversation in most enterprises was theoretical. A committee, a pilot, a Slack channel full of links. Interesting, not urgent.

That posture is expiring. The disruption is no longer showing up as a thought experiment. It is showing up as a competitor's press release, a board question, and a vendor your buyer already pays walking in with an agent bolted onto the renewal.

When I look at where capital concentrated last quarter, it lands on the workflows my buyers own. Help desk. Finance close and AP. Service delivery. Cloud cost. These are the same processes we build agents for. The market is now paying real money to automate them, which tells me the pain is real and the budget is moving.

Growth is still the thing that gets rewarded

One number stuck with me. High-growth software companies are trading around 21x forward revenue. Low-growth ones sit near 2.6x. The spread is not subtle.

The read for founders is simple. The market is not rewarding you for being profitable and sleepy. It is rewarding you for growing into a category that is clearly expanding. AI-native infrastructure is that category. Software with a feature bolted on is not, and the multiples say so.

The quieter warning is for narrow plays

Vertical software lagged the broader index, and the reason given was category-specific disruption risk. In plain terms, buyers worry that a single-purpose tool is the easiest thing for a general model to eat.

I take that as a validation of how we built lowtouch.ai and a warning at the same time. A platform with reach across many workflows reads as durable. A point solution reads as a target. If you run a narrow product, the question is not whether AI reaches your category. It is when, and whether you are the one holding the platform or the one being replaced by it.

What I am actually doing about it

I am not writing this to sound alarmed. I run an agentic platform. A market that is finally paying to put agents into production is the market I have been building for.

But there is a discipline to riding this instead of getting flattened by it. A few things I am holding to.

Land where the pain is provable. Finance, service operations, help desk, cloud cost. The workflows the acquirers just paid for are the workflows my buyers already want fixed. I sell into that, not around it.

Position as infrastructure, not a feature. The valuation gap between AI-native platforms and software-with-AI is not going to close in the laggard's favor. What you are perceived as decides what you are worth.

Respect the CISO. The incumbents buying their way into agents still run on shared clouds and metered credit models. Private, governed, deployed inside the customer's own infrastructure is not a nice-to-have for a regulated buyer. It is the reason the deal closes.

Move fast while the window is open. Incumbents are adapting quickly. The stretch where enterprises feel genuine build-versus-buy anxiety is real, and it will not stay open forever. Getting to production in weeks matters more this year than it did last year. It is also why the old RPA versus agentic automation comparison matters more now than it did when automation was mostly scripts and queues.

The one line I would leave a fellow founder with

AI is real. Not real as in impressive demos. Real as in someone just paid billions to automate the workflow your customer runs every day.

If your ICP looks anything like mine, the disruption already has your buyer's address. The only question left is whether you show up as the partner who helps them get ahead of it, or the vendor they replace when they do.