Sekai
AI mini-app creation platform — text prompt to shareable interactive experience in seconds. 15M+ mini-apps created. 1hr+ average daily engagement. $20M Series A co-led by Khosla Ventures and Connect Ventures. The thesis is right. The retention curve is the one thing the metrics can't yet prove.
What Sekai is building
Sekai is building the creation layer that passive content platforms never offered. Anyone can turn a text prompt into a working, shareable mini-app in seconds — no code, no friction, no skill barrier. The resulting app can be played, remixed, and redistributed. Every remix is simultaneously a distribution event and a creation event.
The founder is Lucky Zhang — two prior exits across Apple and ByteDance, both in consumer media. The pivot from anime fan-fiction platform to general mini-app creation was the most important decision in Sekai's history. It required killing a product that was working in favor of a larger thesis enabled by improving AI coding models. That's constraint awareness at the product level. It's the kind of judgment that doesn't show up in a pitch deck.
A $20M Series A closed June 1, 2026 — co-led by Khosla Ventures and Connect Ventures, the latter backed by CAA and NEA. Total raised approximately $30M. The CAA angle is underpriced in most reads of this deal.
The constraint
The constraint is not distribution. TikTok Minis is a symptom of the real constraint, not the constraint itself. TikTok only becomes existential if Sekai's moat is the creation mechanic alone. If the moat is the community, the remix culture, and the social graph built around creation-as-identity — TikTok building a tab doesn't kill Sekai any more than Instagram building Reels killed TikTok.
The real constraint is creation depth. Specifically: the creator-to-consumer ratio that is not disclosed and not publicly verifiable. 15M mini-apps created is impressive. But if 80% came from 5% of users, Sekai is a power-user-driven content platform with a passive consumption majority. That's a very different business than "creation is the social act."
The 1hr/day engagement number is doing too much work. Session length measures consumption, not creation. The assumption carrying the most weight in the Sekai thesis is that users spending 1hr/day are creating, remixing, and returning to create again — not just playing other people's apps on an infinite scroll. Until that ratio is disclosed, the engagement metric is necessary but not sufficient.
Mechanism credibility
Three compounding steps make the mechanism credible. First: frictionless creation — text prompt to working mini-app in seconds, eliminating the skill barrier that made software creation inaccessible. Second: remix as viral distribution — every mini-app is forkable, making a remix simultaneously a discovery and a creation event. This is structurally analogous to TikTok's duets mechanic but applied to interactive software — and more defensible because the output is functional, not aesthetic. Third: social identity around creation — the cultural bet that creation itself becomes the status act.
Mechanism credibility is high for the first two steps. The third is the cultural hypothesis that determines whether Sekai is a platform or a feature. The birthday card viral loop — documented in the data pack — is the most important proof point. Organic distribution, natural social occasion driving creation, recipient-to-creator conversion path. That's a real flywheel, not a manufactured one.
The CAA pipeline is the most underappreciated element. Connect Ventures brings CAA's talent infrastructure directly into the cap table. CAA represents celebrities, athletes, and IP owners who could build official branded mini-apps on Sekai. This is a proprietary content pipeline no pure-tech competitor can replicate.
Wedge strength
Sekai is not displacing TikTok for consumption. It's displacing the absence of a creation outlet — the gap between "I have an idea for an app" and "I can build it." The wedge is not "watch less TikTok." It's "now you can build the thing you were imagining." That's displacement of an unfulfilled behavior, not an existing one. Harder to prove in early data, but more durable if the community forms.
The Western mini-app format is genuinely unclaimed. WeChat Mini Programs have 450M+ DAUs. No Western platform has captured that behavior at scale. Sekai is not inventing demand — it's attempting to be the first to capture it in the West.
Traction
| Metric | Figure | DILA read |
|---|---|---|
| Total mini-apps created | 15M+ | Volume signal — high. Quality distribution unknown. |
| Daily mini-apps generated | 200K+ | Creation velocity — strong. Creator concentration unknown. |
| Avg. daily time on app | 1hr+ | Engagement depth — exceptional. Retention curve not disclosed. |
| Time to first creation | Seconds | Friction elimination — validated. |
| Total funding | ~$30M | Khosla + a16z Speedrun is a conviction syndicate. |
Founder assessment
Lucky Zhang is the most credentialed founder in the RTC evaluation universe at this stage. Two prior exits — Apple and ByteDance — give him pattern recognition across the full stack of consumer media. The pivot from anime fan-fiction to general mini-app creation is the most important founder judgment signal. It required killing a product that was working in favor of a larger thesis enabled by improving AI coding models.
The anti-doomscroll framing is a founder judgment call, not a marketing decision. It positions Sekai against the entire passive consumption paradigm at a moment of maximum cultural receptivity. Archetype: Signal Builder. Category-aware, technically credentialed, culturally timed.