Welcome to AI Gamechangers, your regular newsletter which publishes conversations with leaders building practical AI into the games industry.
This week we’re talking to Tommi Koirikivi, co-founder and CEO of Theymes, a Finnish startup reimagining how game studios support their players. Tommi makes a compelling case that player support isn’t a back-office function: it’s one of the richest sources of insight a studio has, and AI is finally making it possible to act on that in real time.
First time? Welcome! We have a great lineup of guests coming up, and our archive is full of smart conversations well worth exploring.
Tommi Koirikivi, Theymes
Tommi Koirikivi is co-founder and CEO of Theymes, a Finnish B2B SaaS company building AI-powered player support software for game studios and publishers. His background is in UX and product design, with previous work for clients including Rovio.
Tommi reached out to us by email in April when his company was named a Webby Awards Honoree in the AI, Customer Experience & CRM category. We soon started chatting and learned how generic customer service tools usually fall short in gaming, and how Theymes is a modern alternative that handles the unique emotional and contextual demands of player support. In the Q&A below, we discuss what a truly AI-native platform looks like.
Top takeaways from this conversation:
Theymes decouples the initial AI assistant interaction from traditional ticket handling entirely, resolving many player issues before a ticket is ever opened.
Theymes customers are seeing 30%–50% reductions in ticket volume while simultaneously improving player satisfaction scores, demonstrating that AI-native support can cut costs without degrading the player experience.
Support data in gaming is one of the richest and most underused sources of player insight available to a studio, and Theymes is building towards making that data directly queryable by product, live ops, and leadership teams.
AI Gamechangers: Please tell us your background. What’s your discipline? When did you first join the games industry, and when did you first start using AI?
Tommi Koirikivi: My background is in design and technology, with a primary focus on user experience. Over the years, I’ve been building services and digital solutions for different industries across the globe, usually with the same underlying goal: helping teams work more efficiently and improving the business metrics around the services they operate.
My first touchpoint with the games industry was over a decade ago, when I helped Rovio with one of their products. Since then, I’ve been especially interested in the operational side of games: how studios support players, understand their communities, and build better services around the game itself.
“One of the biggest changes will be how studios understand player sentiment at scale. Games already have a lot of quantitative analytics. The qualitative layer has been much harder to analyse at scale. AI can help connect those dots”
Tommi Koirikivi
I first experimented with “AI” during the previous decade while building solutions in the financial sector. Of course, compared to today’s transformers and large language models, that earlier wave of AI feels very different. I’ve been actively using current AI and LLM-based tools since 2022, and today AI is an integral part of my work across almost everything: how we build, how we operate, and how we think about what new capabilities we can unlock for our customers.
There wasn’t really one single “aha moment” with Theymes. It was more that from the beginning, we were in a position where we could build everything from the smallest details while considering AI as a core part of the system. That lets you approach the product from a very different angle. AI is not something added later. It affects how you think about workflows, data, automation, user experience, and the overall economics of support.
For readers who haven’t come across Theymes before, can you give us the elevator pitch? What problem are you solving for game studios, and why does player support need a games-specific solution rather than an off-the-shelf customer service tool?
Theymes is an AI-native player relationship management platform for game studios and publishers.
We help gaming companies support their players across in-game, web, email, and community channels, while turning player support into a much more valuable source of insight for the whole studio.
The biggest problem we see is that support data is often locked away inside ticketing tools. Studios are sitting on a huge amount of information about what players are confused by, frustrated about, asking for, or struggling with, but that information rarely reaches product, live ops, design, or leadership teams in a useful format.
At the same time, the ticket volumes in gaming can be extremely high. The economics have to work whether a game has 500 monthly players or hundreds of millions. That means player support cannot rely only on adding more people as the game grows. You need smarter automation, better self-help, better insight, and better routing from the start.
“Player support has some very specific AI challenges. Every game has its own terminology, events, items, currencies, mechanics, and community language. There are also sensitive situations where AI should not pretend to know the answer”
Tommi Koirikivi
Generic customer service tools can handle tickets, but games are not generic customer service environments. Players ask about game mechanics, events, items, progression, bugs, missing purchases, bans, accounts, guilds, and highly game-specific situations. The emotional context is also different. A player may be contacting support at the exact moment they are about to churn.
That is why we believe player support needs a games-specific solution. Theymes is built for that reality: fast, contextual, multilingual, AI-native, and connected to the way game studios actually operate.
You describe Theymes as “AI-native”. Can you unpack that for us? How is a platform built around AI from the ground up different from a traditional support tool that adds AI later?
For us, AI-native means that AI is not a feature bolted onto a legacy ticketing platform. It is part of how we think about the whole product.
When a traditional support tool adds AI later, it usually appears as a chatbot, a suggested reply, or a summary feature. Those can be useful, but the core workflow is still the same: tickets come in, agents process them, and the system is mainly built around managing that queue.
With Theymes, AI is involved much more broadly. It is part of the knowledge base, analytics, translations, insights, player-facing assistance, and agent workflows. It also shapes how we build new features. When AI is part of the architecture from day one, you can create capabilities that would be very hard to retrofit into an older system.
“Humans are still essential. Anything the AI cannot handle should always have a clear path to a human. We do not see AI as replacing human support. We see it as an integral layer in the system that helps teams focus their human attention where it matters most”
Tommi Koirikivi
A good example is what we are launching soon with Ask Theymes. The idea is that a customer can ask things like: “Show me where our knowledge gaps are and update our knowledge page FAQs based on this information,” or “What did our VIP players dislike the most during the last month?” That changes how people interact with support data. Instead of manually digging through tickets, dashboards, and reports, teams can ask direct questions and get actionable answers.
So the difference is not just that we use AI. It is that AI changes the way the product behaves, the way teams access information, and the type of workflows that become possible.
Let’s talk about the AI under the hood. What models, agents or approaches are you using, and how do you handle the unique challenges of player support? Are there specific challenges you need to train for, like in-game terminology, spoiler sensitivity, or player emotions?
We use different models and providers for different tasks. For us, it is important to use the right model for the right task. Some use cases need stronger reasoning, some need speed, some need cost efficiency, and some need a very specific balance between accuracy and business performance.
That balance matters a lot in gaming. If your solution works technically but the economics do not scale, it will not work for studios operating at large player volumes. So we do a lot of testing and optimisation to make sure performance and business metrics stay in a healthy balance.
One thing we do differently is that our system handles many player issues before a ticket ever has to be opened. We have decoupled the initial AI assistant experience from traditional ticket handling, and the results have been very strong. Our customers are seeing 30%-50% reductions in ticket volumes while increasing CSAT.
“Theymes also brings new information to the studio through analytics. Product, design, live ops, and leadership teams can understand what players are asking about. Support should not only be a cost centre. It should become one of the best sources of player insight”
Tommi Koirikivi
Player support has some very specific AI challenges. Every game has its own terminology, events, items, currencies, mechanics, and community language. There are also sensitive situations where AI should not pretend to know the answer. Payments, bans, account issues, angry players, VIP players, and emotionally charged situations need the right escalation paths.
So the model is only one part of the system. The more important question is what information can the AI access, what is it allowed to answer, when should it ask for clarification, and when should it hand over to a human?
In player support, a confident wrong answer is worse than no answer. That is why grounding, guardrails, and escalation logic are essential.
You cover a lot of channels: in-game, web widget, Discord bot, email. How does the AI adapt its tone and approach across those different touchpoints? Are there places where you still need humans in the loop?
Our platform is highly customisable per customer. Studios and publishers can shape the branding and tone of the AI assistant so that it feels consistent with their game and player experience.
For us, consistency is very important. The tone should not feel like a completely different company depending on whether the player is in-game, on the web, in email, or in a community channel. The format of the response may change, but the underlying tone and experience should feel coherent.
The AI is best suited for simple questions and tier-one support situations where the player needs quick information to get back into the game. These are often the situations where speed matters most.
“It has to be clear to players when they are interacting with an AI system. In general, players appreciate fast and accurate answers. Especially in mobile gaming, speed matters a lot. Churn can happen in a few seconds. The key is that AI must know its limits”
Tommi Koirikivi
Humans are still essential. Anything the AI cannot handle should always have a clear path to a human. It is also important that our customers can decide where they want to reduce friction and where they want human involvement. For example, some studios may want to give VIP players a different level of support, or they may want certain categories of tickets to always go to a person.
We do not see AI as replacing human support. We see it as an integral layer in the system that helps teams focus their human attention where it matters most.
Can you share a case study of a studio using Theymes? Where is it being used in action today? How have things improved after Theymes was implemented, and are there any numbers you can put on the improvement?
We have released a few public case studies and have more on the way. Across our customers, the pattern is very consistent. Studios are seeing significant reductions in ticket volumes, usually in the range of 30%-50%, while also increasing CSAT.
That combination is important. Reducing ticket volume alone is not enough if the player experience gets worse. What matters is reducing unnecessary tickets while helping players get faster, better answers.
We work with amazing teams from publishers and studios including Saber Interactive, tinyBuild Games, Metacore, Wooga, and Pixel Federation. The use cases vary, but the underlying value is similar: players get better access to help, support teams receive fewer repetitive tickets, and studios get real-time insight into what players are experiencing.
“It is important to use the right model for the right task. Some use cases need stronger reasoning, some need speed, some need cost efficiency, and some need a very specific balance between accuracy and business performance”
Tommi Koirikivi
One of the most valuable parts for our customers is that Theymes not only improves the support queue. It also brings new information to stakeholders across the studio through analytics. Product, design, live ops, and leadership teams can understand what players are asking about, where knowledge gaps exist, and which issues are creating friction.
That is the bigger shift we are interested in. Support should not only be a cost centre. It should become one of the best sources of player insight inside the studio.
Player trust is a sensitive thing, especially when people realise they’re talking to AI rather than a person. How do you handle transparency around that, and what’s been the player response so far?
It has to be clear to players when they are interacting with an AI system. That is very important.
At the same time, studios and publishers need flexibility in how they shape that experience. The wording, tone, and overall presentation can be configured by the customer, but the player should not feel misled.
In general, players appreciate fast and accurate answers. Especially in mobile gaming, speed matters a lot. Churn can happen in a few seconds. If a player has a simple issue and gets an immediate answer that takes them back into the game, that can be a much better experience than waiting for a traditional ticket response.
The key is that AI must know its limits. It should be easy to hand over to a human when the issue requires it. Players are usually not upset because AI is involved. They are upset when they feel trapped, ignored, or given a wrong answer with no way forward.
Our philosophy is: instant help when possible, human support when needed.
Do you see AI-powered support evolving into something broader, more like community management, or even feeding player sentiment data back into game design? Where does this go next for Theymes?
Yes, definitely. I think player support is becoming much broader than tickets.
Support conversations are one of the richest sources of player insight a studio has, but historically, that information has been difficult to use. It gets trapped in tools, tags, and individual conversations. AI changes that because it can help structure and surface what is happening in real time.
Through Theymes, product, design, and live ops teams can gather important player metrics and qualitative insights much faster. They can understand what players are confused about, which topics are creating friction, what VIP players are experiencing, and where knowledge gaps exist.
“Games are not generic customer service environments. A player may be contacting support at the exact moment they are about to churn. Theymes is built for that reality: fast, contextual, multilingual, AI-native, and connected to the way game studios actually operate”
Tommi Koirikivi
Community management is also very important throughout every stage of gaming. Support and community are not completely separate in the player’s mind. They are both part of the relationship between the player and the game.
We are also taking early steps toward expanding our offering beyond the current gaming support use cases. Our SDK can be implemented with just a few lines of code, which makes it very easy for companies to add player-facing help directly into their experience. It would be a bit silly to ignore that this type of approach could work more broadly.
That said, vertical focus matters. Before expanding too far, we need to be clear about which problems we care about most and where we can create the most value.
In the near term, we are very excited about Ask Theymes. It will fundamentally change how our customers use the software, because it turns support data and workflows into something teams can interact with much more naturally.
You’ve just been named a Webby Awards Honoree. What does that recognition mean for a relatively young company? Does it change how studios perceive what you’re doing?
We’re proud that Theymes has been recognised as a Webby Awards Honoree in the AI, Customer Experience & CRM category.
The Webby Awards are one of the most recognised honours on the internet, and being named an Honoree puts us among the top entries globally in our category. For a small Finnish startup building AI-native player support for gaming companies, that is a meaningful recognition.
It is also a strong signal that what we are building is being noticed beyond the gaming industry. Player support has often been seen as a back-office function, but we believe it is becoming a strategic part of the player experience.
For studios, recognition like this helps create confidence. It shows that the problem we are solving is not just a niche support issue. It is part of a larger shift in how companies use AI to build better relationships with their users.
Finally, zooming out: beyond player support, what other AI-driven disruptions do you think are coming to the games industry that people aren’t paying enough attention to yet?
I think one of the biggest changes will be how studios understand player sentiment at scale.
Games already have a lot of quantitative analytics. Studios can see what players do, where they drop off, what they buy, and how they progress. But support and community channels explain why players feel the way they do. That qualitative layer has been much harder to analyse at scale.
AI can help connect those dots. It can turn messy conversations into structured insight for support, live ops, product, design, and leadership teams.
“AI is not something added later. It affects how you think about workflows, data, automation, user experience, and the overall economics of support”
Tommi Koirikivi
Another important area is live ops responsiveness. Games move fast, but player feedback often moves even faster. AI can help studios detect friction, confusion, and sentiment changes earlier, so they can respond before issues grow into bigger problems.
The third area is making support data useful to many stakeholders inside the studio. Today, too much of that information stays inside support teams. In the future, I think the best studios will use player support and community insight as a continuous feedback loop for improving the game.
That does not mean AI replaces the people making creative or strategic decisions. But it can give those people a much clearer view of what players are experiencing, in real time.
Further down the rabbit hole
Some useful news, views and links to keep you going until next time:
Simon Davis, who spoke with us last summer, has opened a UK studio for his rebranded outfit, wearemighty (formerly Mighty Bear Games), to accelerate R&D on SecretSauce, an AI content platform for agencies and brands.
EVE Online studio CCP has rebranded as Fenris Creations and announced a strategic partnership with Google DeepMind, which has taken a minority stake. The collaboration will focus on AI research and development inside the complex player-driven world of EVE Online.
Korean publisher Pearl Abyss, which acquired CCP in 2018, is now selling Fenris Creations back to its board. Evolutionary biologist Richard Dawkins has waded into the AI consciousness debate, sharing his thoughts on whether large language models like Claude and ChatGPT might be genuinely sentient, and what that would actually mean.
Kohort has raised $7 million in a Series A funding round to build AI agents for mobile user acquisition, with its tools trained on $6bn of historical UA spend.
Unity has launched an open beta of Unity AI for all Unity 6 developers: an in-editor agentic assistant with plan mode, LLM integration, and a rollback system for undoing AI-made changes.
Nvidia CEO Jensen Huang has argued that AI is creating rather than destroying jobs, and that doomer rhetoric risks putting people off engaging with the technology. Of course, his company sells the hardware that powers the AI industry…
ChinaJoy this year (31st July-3rd August 2026) is to be themed “Level Up With AI”. The PGC Summit Shanghai event, taking place just before it (29th July), will also feature a number of talks and panels on AI in games.
More detail has emerged on Roblox Reality, the AI-powered photorealism upscaler. SVP Anupam Singh is keen to stress that AI visuals will be subordinate to creator intent. Players will pay a subscription for the upscaled experience; creators won’t be charged extra.
The Academy of Motion Picture Arts and Sciences has ruled that fully AI-generated performances and machine-written screenplays will not qualify for Oscar eligibility from 2027 onward. AI tools remain permitted in production, but authorship and awards recognition stay tied to humans.
Here’s a great talk from the Helsinki AI Gamechangers Summit a few months ago, in which Tuomas Pelkonen of NORDIA Law gave a summary of “AI, Copyright, & The Law: What Every Game Developer & Creative Professional Needs To Know.”
First-line support feels like the most natural spot to apply AI. Almost a decade ago when I was a support agent, ML was used mainly for sentiment analysis and categorization, and then an automation would fire up to serve the customer a relevant but pre-written message with instructions.
These days any off-the-shelf model with a little bit of context and some prompting magic can both recognize the intent and compose a relevant response, more accurate than any boilerplate could ever be. A dedicated service could really cover the vast majority of assistance requests for most companies.