Agents, enterprise, and software’s new role

I recently read a forward-looking piece by Bruno Fonzi on the future of enterprise software—and it stuck with me.

It stuck with me because it captures something I think is arriving much faster than it seems.

It’s not just that enterprise software is getting smarter: its role inside companies is changing.

For years, digital work followed a simple logic: people do the work; software helps.

It gives you better tools, simplifies part of a process, organizes tasks… but execution stays human. You do the work and software supports you—it’s a tool.

With AI agents, that line really starts to move.

Because when an agent doesn’t just suggest what to do but does it—runs a flow, checks data, coordinates operations, closes tasks autonomously—you’re no longer using a tool.

You’re delegating execution, and the agent uses the tools directly.

That sets off a chain reaction.

It fundamentally changes why a company buys software.

It no longer buys a tool to help people work better. It buys an expectation of outcome—an automaton that can run largely on its own and produce verifiable output.

So it’s natural that how that software is judged and paid for changes too: fewer per-user licenses; more value tied to how much work it produces, with what accuracy, with what real impact.

Then there’s a theme that will only get more central: verticalization.

A generic agent isn’t enough—or rather, to make a system truly work inside a company you need domain knowledge (Model Context Protocol, MCP).

Rules, context, real cases, industry-specific processes.

Whoever builds that vertical intelligence isn’t just shipping an “AI product”—they’re building something close to the operating system of that industry or that vertical.

And inevitably, trust shows up.

If an agent works in your place, you need to see everything: what it did, why, with what quality, where it took risk, where it failed. Traceability, continuous control, and security aren’t add-ons—they become part of the final product.

So we’re not really “leaving” the era of tools—but their role shifts; they become infrastructure.

Agents (and MCP-style stacks) use them on our behalf, and that changes how work is done and how software is sold.

It’s a quiet revolution—but in my view, an irreversible one.

Read the original: The Future of Enterprise Software in an Agentic World by Bruno Fonzi.