AI Agents Are Coming for Every Role

Not to replace you, but to give you super powers

A year ago, I couldn’t write a line of code. I’m a growth person — marketing, funnels, data. Then I joined Kilo and started using an AI coding agent in my IDE. I didn’t learn to code. I learned to manage agents that code.

Since then, I’ve shipped landing pages, internal dashboards, and small automation tools — none of which I could have touched 12 months ago. It fundamentally changed how I work. And I’m convinced that what happened to me is about to happen to almost every knowledge worker.

Developers were first. Everyone else is next now.

Over the past year at Kilo, we watched millions of developers adopt coding agents. Many tried a coding agent, hit the rough edges, shrugged it off. A few months later they tried again. The models and tooling had improved, and suddenly the skeptics were using agents daily.

That curve is now repeating for every other role.

But the developer wave revealed something bigger than a productivity gain. The role itself changed. Developers stopped being the ones who write the code and became the ones who direct it. That’s not a subtle shift. It’s a different job. And it’s coming for every knowledge worker next.

Part of what’s accelerating that shift is OpenClaw. Where early coding agents were built for developers living inside a code editor, OpenClaw connects to email, Slack, calendars, and documents. It’s not a coding tool. It’s an orchestration layer for any kind of work. That’s the bridge from software teams to everyone else. KiloClaw is the hosted version. No server setup, no API key juggling, no maintenance. Just the orchestration layer, ready to use.

What it looks like in production

Last week I ran into Anelya Grant, Chief Product Officer at JustPaid, a Y Combinator-backed fintech startup. Her team replaced most of their engineering workflow with seven AI agents, built on OpenClaw and coding agents, each with a defined role, running 24/7. In one month they shipped 10 major features. The human engineers became directors.

But engineering teams aren’t the only ones paying attention.

Pedro Franceschi is 29 years old and CEO of Brex, which was acquired by Capital One for $5.15 billion. He’s decomposed his own job using OpenClaw. A signal ingestion pipeline screens his email, Slack, Google Docs, and WhatsApp, filtering everything through his priorities and the 25 people he cares most about. Granola runs on every meeting, feeds transcripts into the pipeline, and auto-generates action items. For each to-do, the system pulls context from the original meeting and drafts the follow-up. Slack message, email, or text. Pedro just has to click approve.

He also built a virtual recruiter named “Jim” who lives in Slack with his own email address and taught himself to screen fabricated resumes, without anyone explicitly coding that capability. And a security layer called “Crab Trap” intercepts all agent network traffic through an LLM proxy: a second AI monitoring the first in real time.

This is a CEO offloading the cognitive overhead of running a company. And Pedro is just one person. People are already taking this further, trying to run entire engineering teams and companies on agents. Frameworks like Paperclip and Gas Town have emerged specifically for that: orchestrating teams of agents, assigning roles, coordinating work in parallel, and keeping humans in an oversight position rather than an execution one.

The gap between chaos and leverage

Success with agents isn’t just about model quality, though that matters. It’s about how well you can isolate a task, delegate it clearly, and review the output. The skill that matters isn’t prompting. It’s the ability to break work down, direct it precisely, and know what good output looks like. That’s a human skill. Agents amplify it. Here’s a collection of use cases across different roles. Copy any of them straight into KiloClaw to start learning by doing.

The other thing that separates the people who get results from the ones who don’t is simple: they keep experimenting. Many engineers dismissed coding agents the first time around for exactly this reason. The ones who succeeded didn’t wait for the technology to be perfect. They kept experimenting, iterated quickly, and got better at using it through doing.

Real work is also messy and noisy in ways a vague instruction doesn’t always capture. That’s why OpenClaw doesn’t just respond to requests. It learns your preferences over time and acts proactively on your behalf. With that kind of access comes real responsibility. It’s why we invest heavily in security and think carefully about what tools and access each agent actually needs to do its job.

The shift that’s already underway

The people who adapt will look a lot like I do today: not technical in the traditional sense, but capable of shipping things. Not doing all the work, but responsible for its quality and direction. I watched this happen to developers over 12 months. I experienced it myself. It’s coming for every role that runs on thinking, writing, deciding, and communicating, which is most of them.

Nobody announced the moment developers became orchestrators. It just happened, gradually then all at once. The same shift is underway for every other role.

Exciting times to be learning how to 10x yourself.