Three weeks. Thousands of votes. Fans from twelve countries participating in what I think was genuinely unprecedented — a worldwide election to determine which AI-generated characters would form a virtual idol group.
The first AIdeal formation vote closed, and the community has spoken. Two confirmed members emerged from this election: Airi Kazoku and NEON. In this post, I want to be specific about what happened, what the numbers looked like, and what the process taught us — both the things that worked better than we expected and the things we'll do differently next time.
Who Participated and Where
Participation in AIdeal's first formation vote drew registered fans from across East Asia, Southeast Asia, North America, and parts of Europe. Japan and South Korea represented the largest blocs of participation, which was expected given the natural gravity of the regional idol and VTuber fan communities. What surprised us was the engagement volume from communities in Southeast Asia — particularly Indonesia and the Philippines, where virtual entertainment fan communities have been growing rapidly.
North American participation was lower in absolute numbers but notable for its engagement pattern: North American voters tended to participate more actively in community discussions during the nomination phase and showed stronger support for specific character concepts that aligned with globally-oriented aesthetics. This tracks with what the K-pop fandom research community has described about Western fan communities' preference for artists with crossover appeal.
The geographic spread validated a core assumption we had built the platform around: that an AI virtual idol group could draw genuine international investment if the concept and community infrastructure were designed for global participation from day one, rather than Japan-first with international secondary.
The Candidates and What the Community Decided
Seven candidates entered the formal voting phase after the nomination and shortlisting process. Each had a developed character concept, a visual design direction, a vocal archetype, and a community advocacy thread that had accumulated discussion over the preceding weeks.
Airi Kazoku received the highest total vote count across the three-week election. Her concept — a soprano vocalist with an ethereal, celestial aesthetic — had developed strong community support during the nomination phase, particularly among fans who felt the unit needed a classically-trained-register voice anchor. Her visual direction, which leaned into traditional idol aesthetics while incorporating more abstract geometric elements in the costume design, generated some of the most detailed community discussion of any candidate.
NEON placed second in overall vote totals but first in engagement diversity — her community advocacy thread had participants from the widest range of regional communities of any candidate. Her electronic rap concept positioned her as a deliberate counterpoint to Airi's classical soprano: hard-edged, futuristic, with a visual aesthetic that drew on cyberpunk and global streetwear influences. Some community members argued that her concept was too far from traditional idol aesthetics. The debate itself was one of the most substantive creative conversations the community had, and the fact that NEON won despite it suggests a community appetite for genre range within the unit.
A third candidate — who would eventually become Hoshi after an additional round of community development in the second election — received significant votes in this first election but didn't reach the threshold we had set for the maximum two members per election. We'll come back to how Hoshi's confirmation unfolded in a subsequent post.
What Worked in the Process
The nomination-to-voting pipeline performed largely as designed. The four-week nomination phase generated substantially more developed candidate concepts than we had seen in early internal testing, because community debate and collaborative refinement happened at scale. By the time candidates reached the formal vote, fans weren't evaluating bare concept sketches — they were evaluating characters with substantive community-authored backstory and aesthetic direction.
The twice-weekly intermediate tally updates created the escalating drama we had hoped for. During the second week, NEON overtook Airi in cumulative votes briefly, which generated a significant surge in Airi's advocacy community and a counterresponse from NEON supporters. The back-and-forth created genuine stakes that kept engagement high through the final days of voting, rather than the front-runner dynamic leading to early disengagement.
The post-election transparency report was published within 48 hours of vote close. We shared total counts by candidate, regional participation breakdown, and a brief account of three accounts we had flagged for suspicious activity patterns and excluded from the final count (all three were single accounts, not a coordinated campaign). The community response to that level of transparency was uniformly positive — even fans whose preferred candidates didn't win reported feeling the process was fair.
What We Learned the Hard Way
The nomination phase produced one situation we hadn't fully anticipated: two very similar candidate concepts from different nominators generated a split community response. Fans who loved the underlying concept were divided between two candidates, and the resulting vote split likely cost both candidates a result that either might have achieved with unified support.
We've since introduced a similarity review step in the shortlisting process, where the team flags candidates with significant conceptual overlap and brings the relevant community nominators into a conversation about whether to merge or differentiate. We can't force mergers, but we can facilitate the conversation before community energy solidifies around duplicate entries.
There was also more friction than expected around account verification for fans in regions where standard email verification faces reliability issues. We lost some genuine participation because account creation failed at the verification step. We've since added alternative verification pathways, and we'll publish the specific changes to the community before the next election opens.
What Comes Next
Airi and NEON are confirmed AIdeal members. Their voice models are in active development. Character design is in detailed refinement phase, informed by the community discussion threads that shaped their concepts during the nomination process. Performance debut is on our development roadmap.
The fourth member slot — beyond Hoshi, who was confirmed in the second election — remains open. The next formation vote will use updated rules reflecting everything outlined above. Nomination windows, eligibility requirements, and results publication format have all been revised based on this first cycle's learnings.
I want to close by saying something that I mean without any marketing intent: the community that participated in this first election is why we started Paradigm AI. Not the technology, not the idol format, not the business opportunity. The thing we were trying to build was a place where fans could have genuine creative stakes in an entertainment property they love. Watching that happen at real scale, across twelve countries, generating debates about character aesthetics that were more substantive than anything we could have generated internally — that's the thing we came here to make.
Thank you to everyone who nominated, debated, and voted. AIdeal is yours in the truest sense.
