The Future of Work in UA Is Managing a Team of AI Agents

I can't imagine we'll recognize the world of work a few years from now. I work alongside super smart people at the frontier of agentic AI every day. It feels a bit overwhelming.

When you see firsthand what AI agents actually do, running autonomously, fully aware of the relevant context, working for hours at a time towards a shared objective, you can't stop thinking about what work will look like in a few years.

To be frank, I didn't upload this article to this website myself. I didn't QA everything here, and I didn't do the research alone. The task I had this week to write this piece was tracked and updated without me touching monday.com.

I didn't log into Google Analytics or our CRM this week. But I got a daily performance summary from my agent. I'm way more informed about what's happening with marketing than I was a couple of years ago, when this process depended on humans.

I get every piece of content anyone on Earth writes about mobile games. I know about every game that launched, which creatives are winning in each genre. Every day.

I wake up every morning with most of my mundane stuff ready to go. My agent listened to all my calls, read all my emails, coordinated what needed coordinating. It created everything I need to perform 100x better than I could alone.

But even though I've loved mobile games, UA, and the business of games for over a decade, I'm still just a B2B marketer. You're a UA person, a creative person, or someone on the business side of games reading this.

This technological wave is washing over the world like a tsunami. Most people don't feel it yet. It's not spreading evenly across all walks of work life.

The #1 use of AI agents right now is coding. Tremendous progress happened in the past year or two. The best developers are managing teams of AI agents that write the vast majority of their code.

This exact same thing is going to happen to you, to me, to all of us. More than that, it's going to be so aggressive that there probably won't be a job market for AI-illiterate people. I'm not talking about chatting with some random LLM. That's a toy.

I'm talking about dispatching agents every day that do most of the work you used to do - making you the CEO of your role - managing a little company of You.

What Managing AI Agents Is Going to Look Like in UA

UA for mobile games has always been a brutal role. You usually don't control which game you market. You get a ROAS target, a budget spend target, a bunch of downstream metrics as secondary targets. And you usually get creatives from a different team.

You still get the blame when UA "isn't scaling". Most of your time goes to opening new channels, setting up campaign structures, optimization strategies, measurement, analytics, reporting, bid optimizations. Data analysis, data analysis, data analysis.

And as networks handle more of this, a huge chunk of effort has shifted to creatives and creative optimization.

Think about the entire loop of running and scaling UA for a mobile game. Some parts are handled by adjacent roles: creative strategists, directors, artists, analysts. But every studio has to face each of these parts.

Research

Usually done by creative strategists or directors, but often by UA people too. Researching what's happening in your genre from a creative perspective takes a big chunk of time.

Scrolling through Sensor Tower or AppMagic. Browsing social media feeds and ad libraries, studying competitors' creatives, looking at what's trending. Looking at what the most successful titles are doing in UA creatives, over and over again.

Researching the game itself, connecting to upcoming LiveOps events, trying to land on actionable insights.

Ideation

Then taking all that research and trying to translate it into clear theses, hypotheses, and creative ideas.

Developing these into actual creative briefs, adjusted for each network. And most painfully, given that production capacity is limited, trying to prioritize which ideas to take to production based on... hunch? Some data? When you have limited creative capacity, you're trying to predict what will be successful before the idea even makes it to production.

Production of Videos and Playables

Taking these briefs to an in-house team, agencies, or anyone who can execute them at the scale required. Few industries understand how many creatives it takes to market a single mobile game hit.

And waiting. And waiting some more. Checking emails. Waiting. Saying in the weekly meeting that you're still waiting for more creatives for this and that network.

Eventually you get creatives in the hands of the UA team.

Deploying Them to Networks

Spending time uploading creatives to campaigns. Properly named with unique conventions, technically correct so they get approved. Thinking about how to craft the right creative sets to maximize performance.

Observing and Analyzing

Giving it time. Watching data come back through your MMP. Analyzing campaigns at the creative level, sifting through hundreds or thousands of rows. What works, what doesn't. Which concepts show promise. Which are dead in the water. And hopefully, what's the winner.

Then packaging this into insights that feed back to the creative teams. Better ideas. A higher hit rate.

And repeating forever. Or until you retire, or move to a new game company where you repeat that again in a slightly different way.

AI Agents Are Going to Run This Entire Process

Software developers dispatch coding agents to research codebases, plan features, write specs, plan sprints, actually code, and QA it all autonomously. They hand over working features ready for human review.

It's not perfect yet. But it's getting there damn fast.

How is this going to affect UA as a craft?

You Manage the Agents. They Run the UA Loop.

Instead of researching competitors and trying to fit a ton of data into a human mind, agents crawl the entire creative space of your genre. They see what's working at scale, sifting through patterns, structures, and shared traits of winning creatives.

Because agents aren't limited by time or energy, they explore the vast space of creative possibilities. They help you recombine mechanics from completely different genres with similar play styles. And stumble on the next "save the king" concept.

You wake up. A report of what's new from yesterday. Everything ready for you to make decisions.

These agents turn research into detailed creative briefs, with game design and UA expertise baked in.

On the way to the office you swipe through the ideas on your phone. Nod. Approve most of them, reject what doesn't look right. Go grab your coffee.

While you do that, agents already passed the briefs to another team of agents. They're creating videos and coding playables from scratch. True to your brand. Fun game experiences with solid level design and game juice.

After you finish your Cortado (or Matcha, if people still drink that monstrosity), you sit at your desk. You get a ping.

The agents just messaged you on Slack with 30 new playable ads. They've already QA'd them, packaged everything correctly for every network. You play them, have fun. You like 25, text the agents to upload them to the networks. You ask for changes on the other five. Thirty minutes later, a ping. Done. Fixed according to your feedback.

New playables are live on AppLovin, Unity, Google. You focus on strategy. Thinking about how to find new audiences, leaning into your taste in UA and games. You collaborate with another team of agents, sending them to analyze an idea you had while playing a game over lunch.

They come back with the analysis. You also got an email from your agents on last week's campaign performance. There's a new winner concept from the last batch of creatives. It's scaling in spend aggressively. You smile and email back: hey, send this report to the team.

If That's the Future, What Do You Do About It?

I believe this is the future because I see it day in and day out. Still early innings. Barely started. But it's happening fast.

We're building exactly this at Sett. We started with creative production, more recently focused on playables. But we're building the future of UA work through agents.

More on that real soon. With examples and everything.

In the meantime, get well versed in what agents are and how to work with them. I'm working on a 1-hour crash course on agents for UA people. Coming out soon, for free, here on the Sett website.

Here are some things you can explore right now:

AI Agent Tools Worth Knowing

👉 Claude Cowork - Anthropic's agentic AI system for knowledge work. Unlike regular chat, Cowork lets Claude complete work on its own. Describe the outcome and cadence, and it takes action and keeps you informed. It handles multi-step tasks like organizing folders, pulling metrics, drafting reports. Not just describing how. Actually doing it.

👉 Scheduled Tasks (inside Cowork) - Have Claude check your email every morning, pull metrics, or run your weekly digest. Define the cadence once, Claude handles it from there. This is the "morning briefing" scenario I described above. Live today.

👉 Claude Dispatch - Released March 17, 2026. Send instructions to Claude from your phone while AI executes tasks on your desktop. Claude as a remote employee, completing assigned work independently. Dispatch is part of Cowork.

👉 Computer Use in Cowork - Released March 23, 2026, still a research preview. Claude can control the Mac desktop, open apps, navigate the browser, fill out spreadsheets. When no connector exists for a service, it falls back to operating the keyboard and mouse directly.

👉 Claude Code Desktop - Claude Code in the desktop app. Run multiple agent sessions in parallel: one fixes bugs, another researches GitHub, a third updates docs. This is the developer world I referenced. Managing a team of agents in parallel.

Get AI Literate

👉 AI Fluency: Framework & Foundations - Anthropic's own course on AI literacy. Something for everyone, whether you're new to Claude or a seasoned practitioner.

I want you to set an objective: become AI literate. Not buzzword literate. Actually literate in working with agents. There are tons of resources to get started with, and I'm working on specific ones for game UA.

I challenge you to spin up Claude Code and start playing with it this weekend. Or next week, if you have nice plans. Life is more than AI and games.