Why I'm Excited About This Moment in Games: AI-Enabled Leverage in a Changing Industry

The old operating model

"This is hugely expensive and incredibly complicated" I thought to myself. It was 2021, and I had just completed running an audit of a data organization that I had recently taken ownership of in a new role. The team had been working for several years on a cloud migration that was stalled out and failing to deliver value, and I was brought in to turn the project around. The current state of affairs included multiple separate teams under different management trees, along with a multi-million dollar per year outside consulting group who was hands-on-keyboard building. Data flows criss-crossed between teams and infrastructure like a crazy pipes screensaver.

That was just 5 years ago, but already the available tools and processes for technological transformation at an organization have completely shifted. Now we have so much implementation power available at previously unheard of cost. If you wanted to deliver technology into a business that was outside your internal team's expertise at that time you essentially had two options:

1. Take one of your best team members (or a whole team) off of regular business projects and tell them "go evaluate this new piece of technology". They would know the business context, but not the tech. They could spend weeks working with and evaluating the tech, and after that they might have some idea of how to deploy it, but no experience at scale or in production.
2. Bring in an outside expert. They would come in with hard-won opinions from running with the tech at scale at other organizations. But they wouldn't know the ins and outs of your business, and they wouldn't know very well how the solutions that worked for other clients would pan out for your use cases.

In either case, there would be a period of weeks or months where the individual or team involved would be trying to patch the holes in their context, either the technology context or the business context.

What AI agents changed

Today, we have AI agents - LLMs paired with tools and execution environments. They can consume business context incredibly rapidly. Their training data includes many of the pitfalls and weaknesses that experienced operators have encountered, and their research modes can uncover more specific context on these areas very rapidly. You can spin up POCs with agents in hours, work that would have taken weeks of software engineering time. The compression between well-defined business needs and technical solutions is real, and it continues to accelerate.

"The compression between well-defined business needs and technical solutions is real, and it continues to accelerate."

What this looks like at Plan A

At my current role I work across the technical and business sides of Plan A Games. We're a focused startup building a fresh take on UA funding, and (as with any startup) I am regularly wearing many hats. I move between technical strategy, business liaison, game performance modeling, engineering leadership, and team management. The amount of throughput that I've been able to achieve with the new LLM tooling would be previously unachievable, and it continues to go up as I spend more time with the tools, and the models and harnesses get more powerful. The breadth of value I can bring also continues to go up, as I can execute high level ideas that would have previously been handed off to a specialist.

The shift of what's possible is inspiring. When we were establishing the Plan A Games business presence we initially were fielding $10k-$20k proposals for a custom website. We built it in-house with LLM assisted workflows for a fraction of that cost with relatively little dedicated time. And (bias alert) the finished product looks very high quality - better than the portfolios of some web dev shops. When I want to provide direction to someone on my team I no longer have to be limited to high level ideas that may or may not translate well. Now I can externalize the rough implementation of my ideas to a custom agent and have it prove or disprove my thesis, and then provide an already-validated idea with a POC in place to the team. I am seeing in real time 1) the cost of software implementation driving down, 2) the size of communication gaps narrowing, 3) the iteration speed to materialize and refine an idea plummeting.

For example, recently I was exploring a question around ROAS prediction methods. I spent about an hour directing an agent to implement a quick growth ratio model and run backtesting scenarios. With high level direction it was able to explore my intuition around predictive power in relation to cohort aggregation sizes and UA performance dimensionality. I was able to hand off my findings and the methodology to my team to get them on the right track for implementing a product feature in our data platform. In previous operating modes, it could have easily taken several days of back and forth to communicate my intuition, explain how it would work with our data, have a data scientist manually build a framework, and then run multiple tests with iterative feedback to identify what was working and what wasn't.

The compounding effect

All that and I feel like we are just scratching the surface on the new wave of operating models. Not only will each person be able to do more, there's a significant compounding effect that happens when you consider efficiency of an organization as individuals move up the adoption curve. As the number of people required to run an org reduces, communication friction decreases, layers of hierarchy are eliminated, and ideas encounter less dilution as they go through the execution process. There are still a lot of open questions around how exactly efficiency gains scale. Identifying effective operating models that can take advantage of throughput gains while reducing the impact of agent weaknesses is an active area of discovery across the industry.

Why this matters for games

Now extrapolating my experiences and the trends I see to the broader industry, I believe we are on a trajectory towards exciting outcomes. The team sizes that are required to deliver software functions for building and operating a tech company are reducing. Plan A Games is positioned at a powerful leverage point in the arc of the gaming industry, because I believe we are poised to see a new wave of innovation and creativity. There has been a significant concentration of mobile products through a wave of high-cost M&A, and VC appetite for funding large studios or long runways is quite reduced at this point in time. Along with those financial constraints has come an expected narrowing of what types of games are making it from concept to launched product. But I am seeing dev studios deep into the adoption curve of agent technology achieving outputs traditionally associated with much larger teams. The role for Plan A Games in this shift is to partner with the studios that are delivering great products to players. Better, more innovative player experiences in the marketplace are the spark, and Plan A's UA capital is the fuel.

"Better, more innovative player experiences in the marketplace are the spark, and Plan A's UA capital is the fuel."

I started my time in the industry working in a 70 person company that built and operated a highly successful mobile product. Now, I'm seeing builders creating similar product quality with a fraction of the people. How many more games can be built now? How much less capital intensive is it to build games? How many of those developers that previously would have been reliant on large VC budgets and the boards that they bring are now going to be creating games with relatively unencumbered independent financing? We may not have the hard numbers to answer those questions objectively, but the trend looks clear: more games, more creativity, and more innovation entering the space. I am energized by the future of the industry, both as a business operator and as a lifelong gamer.

Back to top