ZeniMax turns 30-minute safety investigations into two-second reviews

A game launch doesn’t wait for a support team to catch up. Neither do the players who show up the moment it goes live.

ZeniMax is the parent company behind several video game studios, including the team behind the massively multiplayer game Elder Scrolls Online, and it handles gamer support across account access, subscriptions, game progress, and player conduct. Most of that support runs through email. In the weeks before Elder Scrolls Online launched, the team was fielding about 30 questions a week. Launch week brought about a million interactions, arriving faster than any team could answer one at a time.

To keep up, ZeniMax built a way to answer high volumes of similar tickets quickly and accurately by email, backed by thousands of approved knowledge articles. It worked well enough that customers started choosing email over phone and chat on their own. Eventually, ZeniMax turned both of those channels off. Templates and scripts kept about 70% of email volume moving without a live agent needing to step in.

That system had a ceiling, though. Two problems kept the hardest tickets hard:

• Player conduct complaints carried real legal weight and real research time. About half of every ticket queue was one player reporting another, for harassment, cheating, or another rule violation. Each one needed careful, deliberate handling given the legal exposure involved, and an agent typically had to search three to five separate account systems by hand just to understand the relationship between two players before making a call. A complex case could take half an hour of research. A cheating investigation could run into hours.
• Every game ran on its own proprietary technology, with no shared way to search it. ZeniMax supports multiple titles, and not every one had automatic log retrieval built for its agents. For those titles, agents still ran searches manually, one game, one logging system, one query at a time, with no standard way to pull the answer faster.

Scripted email had solved the easy tickets. It hadn’t touched the ones where an agent still had to read, judge, and research a complaint before doing anything about it.

ZeniMax’s own engineering team used Quiq AI Studio to layer two AI use cases on top of its existing email operation, rather than replacing what already worked. The shift wasn’t about routing more tickets to a bot. It was about giving agents a head start on the tickets that took the most time to work, the ones where the answer wasn’t a template, or where the ticket needed a judgment call about another player’s account.

• Match emails to the exact answer, not the closest one. Every inbound email is automatically read and categorized, then matched against ZeniMax’s own library of thousands of approved knowledge articles. Instead of returning a generic response, the AI answers from the specific sub-section relevant to that player’s issue, using language ZeniMax had already written and approved.
• Flag player conduct complaints as they arrive. A separate AI model reviews what players say about each other and labels a ticket for safety review when a complaint could lead to disciplinary action, since these tickets carry legal exposure that requires careful handling.
• Pull account history and evidence automatically. For flagged safety tickets, the AI checks the accused player’s account history and gathers the relevant evidence, work that used to require an agent to run four to six separate queries across ZeniMax’s account systems just to understand the relationship between two players.
• Generate a recommendation, not just a summary. The AI doesn’t stop at organizing evidence. It produces a specific recommendation, such as a suggested ban, along with the reasoning behind it and what happens next, so an agent’s job becomes reviewing a case instead of building one.
• Build and maintain it in-house. ZeniMax’s developers build, test, and maintain these AI use cases directly in Quiq AI Studio, with Quiq’s engineering team supporting through regular working sessions rather than owning the build themselves.

From a player’s perspective, the shift shows up as a faster, more specific answer. From an agent’s perspective, it shows up as a case that arrives already built.

Getting unstuck without a phone call. A player emails ZeniMax after getting locked out of their Elder Scrolls Online account. The AI reads the message, matches it to the specific troubleshooting article for account access issues, and sends back a direct, accurate answer. No phone tree, no queue, no waiting for a callback. If that answer solves it, the ticket closes there.

Reviewing a case instead of building one. A player emails to report someone using abusive language in a match. The AI categorizes it as a safety complaint, pulls the accused player’s account history and prior violations, and puts together the evidence, before a human agent ever opens the ticket. The agent isn’t starting an investigation. They’re reviewing one that’s already assembled, along with a recommended action, and deciding whether to approve it.

These aren’t hypothetical scenarios. They’re why a ticket that used to take an agent half an hour of research now takes about two seconds to review.

For ZeniMax’s support agents, the job changed underneath the ticket. On the email side, agents used to match a customer’s issue to the closest approved response by hand. Now that matching happens automatically, so agents step in mainly when a ticket doesn’t fit a known answer.

On the safety side, agents no longer open a fresh investigation for every player conduct complaint. They aren’t logging into three, four, five account systems to compare two players’ histories or figure out whether a report is retaliation or a real violation. That work is already done and waiting by the time the ticket reaches their queue.

Agents review, decide, and act, instead of researching, deciding, and acting. That shift shows up most on the hardest tickets, the ones ZeniMax used to measure in hours, not minutes, where the team is now confirming a case instead of building one from nothing.

ZeniMax’s support operation was already efficient before this AI layer went in. The change isn’t a jump in email volume handled, it’s what happens to the tickets that used to require the most judgment. Player conduct complaints, roughly half of all tickets, no longer start from a blank account history. Agents get a finished case to review instead of a queue of separate systems to search.

ZeniMax’s own engineering team continues to build new AI use cases directly in Quiq AI Studio, with additional support workflows in active development.

Key outcomes:

• Contextual answers, not generic ones. The AI matches each email to the exact sub-section of ZeniMax’s knowledge base, not a one-size-fits-all template.
• Evidence assembled automatically. Account history and violation evidence for player conduct tickets are compiled before an agent opens the case, replacing manual searches across multiple systems.
• Built and run in-house. ZeniMax’s own developers build and maintain its AI use cases directly in Quiq AI Studio.
• A multi-year AI partnership still expanding. ZeniMax has layered new AI use cases onto its support operation over time, with more in development.