How InnoGames used AI; not for growth but long-tail profitability
InnoGames' Sunrise Village is in long-term decline, but it's now profitable thanks to the use of AI.
In an industry desperate for real-world examples of AI use in live games, InnoGames' enthusiasm to talk about its work on mobile game Sunrise Village has provided a juicy opportunity.
Starting with a GamesIndustry.biz article, then compounded more recently by a Deconstructor of Fun podcast and case study, Sunrise Village has become a surprising poster child for both the limitations and the opportunities available when using AI for legacy content.
Launched in 2022, the game had plateaued by early 2025 and began losing players. Downloads halved. Revenue dropped 20%. InnoGames had a decision to make. It decided to leverage AI.
The studio does not claim the game was resurrected. Instead, its account frames AI's contribution as extending the title's viable commercial life at a lower operating cost.
The team assigned to Sunrise Village was reduced from 25 people to three. The remaining 22 staff were not laid off but reassigned to a new InnoGames title, Cozy Coast. The studio describes the financial outcome for Sunrise Village as moving from "deep red to green" — a reference to cost structure rather than revenue growth.
Downloads and revenue continued to decline after the AI tools were introduced, according to the report, but at a slower rate than before.
The case study identifies the specific bottleneck that prompted AI intervention: a designer was manually entering design definitions into Confluence, exporting the data to Excel, converting it to CSV or JSON format, and then importing it into the Unity game engine. This was a four-step manual process spanning multiple tools.
An engineer at the studio identified the inefficiency and built a custom tool, reportedly in around an hour, to automate the pipeline. The report frames this as representative of where AI delivered value at InnoGames, not in terms of content generation, level design, or other creative functions, but in eliminating repetitive data-handling tasks in the production pipeline.
The case study also notes that InnoGames continued using OpenAI's GPT-4o model for approximately 18 months, despite the release of newer models during that period. The studio prioritized pipeline stability over adopting newer models, according to the report, reflecting a preference for predictable performance in tools running semi-autonomously in production rather than pursuing incremental capability gains.
Deconstructor of Fun frames the InnoGames account as a departure from more common industry narratives in which AI adoption is credited with reversing a game's decline or driving significant revenue growth.
The report characterizes those narratives as marketing-oriented and often lacking in operational detail, and positions the Sunrise Village case as a more representative account of how AI tools are being used in live-service game production: to automate specific, previously manual workflows, and to help studios reallocate staff toward newer projects as older titles are run down with reduced headcount.
The report does not quantify the cost savings generated by the automation, nor does it specify how much longer Sunrise Village's commercial life was extended as a result. Neither does it indicate whether InnoGames plans to apply the same custom tooling, or similar AI-driven pipeline fixes, to other titles in its portfolio.
Yet the account is notable within games industry AI coverage for its specificity and its refusal to overstate outcomes. Rather than presenting AI as a driver of growth, the InnoGames case describes it as a cost-management tool applied to a title already in decline, with the freed-up headcount redirected to a different project.
Industry observers tracking AI adoption may find the account useful as a benchmark for evaluating similar claims, given the tendency for AI-related case studies to omit details on baseline performance, headcount changes, or the specific nature of the workflows automated.
