Gaming in the Era of Agentic AI: Insights from EPAM’s Devcom 2025 Panel
At Devcom 2025, gaming leaders from Xbox, KRAFTON, Xsolla and PlayStation joined EPAM’s Head of Gaming Vitalii Vashchuk, for a conversation as forward-looking as it was grounded in the realities of today: What does Agentic AI truly mean for the gaming industry, and how do we ensure it serves both creativity and players?
Over the course of an hour, the panel went beyond the hype to explore how AI is transforming every layer of gaming - from how worlds are built to how development teams work daily. It was part prediction, part practical roadmap and part rallying cry to build the future together.
You can watch the full Devcom 2025 panel discussion now to hear every insight, debate and bold prediction first-hand - and explore the conversation in its entirety.
While the perspectives were diverse, the five industry leaders on the panel shared a clear sense of where the industry is headed. Below, we’ve highlighted some of the key themes and takeaways from the session - ideas that resonated strongly and offer a glimpse into gaming’s AI‑powered future.
From Automating Repetition to Powering Innovation
One of the first points of agreement was that AI’s role isn’t to replace people, but to unlock their potential. Thomas Ko explained it in clear terms:
“AI is not here to replace developers but to support them in becoming better developers. Corporates need to make that decision. For us, we made it very clear: we’re never going to use AI to replace developers. Instead, we use it to make them more productive. For example, if something used to take three or four years of development, we aim to shorten that and get to market much quicker. That’s the impact we’re trying to achieve.”
By eliminating the burden of repetitive systems and live ops maintenance, studios can reallocate resources toward developing richer narratives, more ambitious worlds and innovative gameplay mechanics. This shift lets creative teams focus on the human spark that makes great games memorable.
Worlds That Truly Feel Alive
What happens when game AI stops following rigid scripts? The panelists painted a vivid picture of emergent NPCs and game systems that react in real time to players’ choices, moods and play styles.
Moin Moinuddin was especially animated:
“I think a fundamental thing that will change is moving from scripted NPCs to emergent NPCs. Today, NPCs are scripted, but tomorrow they will be much more emergent, adapting dynamically based on the game, the user, and the experience. [...] Lastly, I would say hyper-personalization is key. While personalization today tends to be static, tomorrow, with real-time data, it will become far more dynamic.”
Haiyan Zhang shared that Xbox is already experimenting with NPC helpers that offer context-sensitive tips while you play, reducing friction in the moment instead of forcing players to break immersion:
“Having Agentic AI that supports players without distracting them is key. Everybody has those moments where they get stuck and turn to Google or YouTube for help. [...] AI can step in during those moments and help players get back on track right away.”
In this future, no two players will experience the same story.
Tackling the Hard Questions
Of course, innovation comes with hurdles. Cost topped the list. Real-time AI models can carry token and infrastructure expenses that make them prohibitive for wide-scale deployment. As Thomas Ko warned:
“The cost is ridiculous at this point. Unless the cost of GPUs or LLMs comes down, it’s not possible - unless games are priced at $1,000 each, which simply won’t happen. So, cost is prohibitive.”
Trust and IP rights are an equally pressing issue. Moin Moinuddin noted that while many studios are already experimenting behind closed doors, few want to publicize their use of AI due to unresolved questions over ownership of AI-generated content. Until the industry establishes clear guardrails, public adoption will remain cautious.
Practical Advice for Studios Taking the First Step
For all the big ideas shared, the panel also offered pragmatic guidance. Luis Villegas suggested:
“If you haven’t used AI yet and you’re trying to understand how to bring AI to your studio and your team, the biggest piece of advice I have is use it in a way that is non-controversial. Get a win that allows you to show your colleagues how effective AI can be. [...] We started having a bunch of wins in the operational space without major investments and showed it to the rest of the studio - and that kicked off a lot of goodwill and interest.”
Meanwhile, Haiyan Zhang encouraged experimenting with different agentic coding tools and models - even if it’s just to handle documentation or routine code rewrites - to develop an “AI recipe” that works for your unique studio setup.
A Future That Belongs to the Player
As the discussion wrapped up, the tone shifted from strategy to vision. Luis Villegas offered a line that seemed to hang in the air after the session ended:
“Tomorrow, the game you play will be just your game, personalized with the parameters you choose. Nobody else will have the same experience.”
That single thought captured the promise at the heart of Agentic AI: a future where games aren’t static products, but living, breathing worlds shaped in part by the player themselves. A game that will be about the players choice, journey and story.
You can now watch the full Devcom 2025 panel recording on demand and experience the conversation as it happened - insights, debates and visionary ideas from some of the brightest minds in gaming AI.
Watch now and explore how your team can responsibly harness AI to create, innovate and connect like never before.
