{"rewrite":{"id":"r_8af4ea50cd761c37452acf35","clusterId":"c_1bf3e0da7e8b244bc988a2d9","slug":"embark-studios-details-machine-learning-lessons-from-arc-raiders-and-the-finals","model":"deepseek-v4-flash","headline":"Embark Studios Details Machine Learning Lessons From Arc Raiders and The Finals","summary":"At NDC26, Embark Studios machine learning lead Martin detailed how the studio uses ML for recommendations, quest generation, and enemy behavior. A key takeaway was that a simple EASE recommendation model increased purchasers by 400% in The Finals, but organizational hurdles nearly killed the project until an engineer physically moved desks to the commercial team.","whyItMatters":"The talk shows that for Embark, the hardest part of deploying machine learning in games is not the algorithm but the organizational structure around it.","webCardHtml":"\u003cp\u003eEmbark Studios machine learning lead Martin gave a session at NEXON\u0026#39;s NDC26 conference covering three practical ML applications across The Finals and the upcoming Arc Raiders. The first case study focused on The Finals store, where a simple EASE recommendation model-just four lines of Python-was tested in the For You section. The result was a 400% increase in purchasers and a significant rise in in-game currency spending.\u003c/p\u003e\u003cp\u003eDespite the algorithm\u0026#39;s success, the project nearly failed due to organizational friction. The ML team was isolated from the commercial team, leading to misaligned baselines and a false initial analysis showing no significant improvement. The fix was mundane: the ML engineer moved his desk to sit with the commercial team. Communication overhead vanished, tests passed, and the model shipped.\u003c/p\u003e\u003cp\u003eMartin also discussed using LLMs analytically-not generatively-to validate quest graph structures, since LLMs can read every permutation of quest order without fatigue. The third example, Arc Raiders robots, was mentioned but not detailed in the report.\u003c/p\u003e","blueskyPost":"Embark's ML lead at NDC26: a 4-line Python recommendation model boosted The Finals store purchasers by 400%, but the project almost died because the ML team sat in a different office. The fix? Move desks.","twitterPost":"Embark's ML lead at NDC26: a 4-line Python recommendation model boosted The Finals store purchasers by 400%, but the project almost died because the ML team sat in a different office. The fix? Move desks.","threadsPost":null,"newsletterBlurb":"At NDC26, Embark Studios ML lead Martin revealed that a simple EASE recommendation model increased The Finals store purchasers by 400%, but organizational isolation nearly killed the project. The solution was moving an engineer to sit with the commercial team. He also discussed using LLMs analytically for quest validation.","attributionJson":"[{\"source\":\"4Gamer.net\",\"url\":\"https://www.4gamer.net/games/652/G065211/20260616016/\",\"title\":\"「ARC Raiders」のロボットはどう動いているのか。Embark Studiosが語った機械学習の実例と，「組織とツール」という泥臭い教訓をレポート［NDC26］\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":5798,"outputTokens":644,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1781612759,"createdAt":"2026-06-16T12:16:00.000Z","publishedAt":"2026-06-16T12:19:57.000Z","updatedAt":"2026-06-16T12:19:57.000Z"},"cluster":{"id":"c_1bf3e0da7e8b244bc988a2d9","canonicalTitle":"「ARC Raiders」のロボットはどう動いているのか。Embark Studiosが語った機械学習の実例と，「組織とツール」という泥臭い教訓をレポート［NDC26］","representativeArticleId":"a_9a00d48d97f44f7f648899db","sourceCount":1,"writtenSourceCount":1,"writeAttempts":0,"isSolo":false,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[\"ARC Raiders\",\"THE FINALS\"],\"studios\":[\"Embark Studios\"],\"people\":[\"マーティン\"],\"type\":\"news\",\"domain\":\"games\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-06-16T11:43:11.000Z","lastSeenAt":"2026-06-17T05:40:06.000Z","updatedAt":"2026-06-17T05:42:21.000Z"},"attribution":[{"source":"4Gamer.net","url":"https://www.4gamer.net/games/609/G060942/20260617013/","title":"1600万本のヒット作「ARC Raiders」の3Dコンテンツは，約20人で作られた。Embark Studiosのツール戦略とは［NDC26］"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":["ARC Raiders","THE FINALS"],"studios":["Embark Studios"],"people":["マーティン"],"type":"news","domain":"games","is_roundup":false},"keyFacts":["A simple EASE recommendation model, written in four lines of Python, increased purchasers by 400% in The Finals store.","The ML project nearly failed because the ML team was isolated from the commercial team, leading to misaligned baselines and a false initial analysis showing no improvement.","The fix was an ML engineer physically moving his desk to sit with the commercial team, which eliminated communication overhead and allowed the model to ship.","Embark Studios uses LLMs analytically to validate quest graph structures, as LLMs can read every permutation of quest order without fatigue.","Embark Studios machine learning lead Martin gave a session at NEXON's NDC26 conference covering three ML applications across The Finals and Arc Raiders."]}