AI-native infrastructure for live audience participation.
PLLAY turns any live audience into active participants — gaming first — with real engagement and credit payouts, and runs the whole operation on a fleet of twelve autonomous AI agents. Skill-weighted prediction contests are live today; the agent layer that settles them, polices fraud, resolves disputes, and supports players is how a small team runs a money-adjacent platform.
Live gaming generates enormous engagement and very little of it flows back to the people creating the moments or the fans showing up for them. Attention gets captured; value rarely gets shared.
Gaming was the wedge, not the boundary. The same rails now run on music streams — DJs opened the first non-gaming vertical with track boosts, setlist bingo, and music trivia — and the engine underneath doesn't care what's on screen, only that a crowd is watching it live.
PLLAY exists to close that gap with skill, not chance. Outcomes are driven by what fans actually know about the game and the creator — and payouts are weighted by a track record of being right. It's a contest of judgment, not a wager against the house.
And we built it AI-native from day one. The work that usually needs a back office — settling contests, catching fraud, ruling on disputes, re-checking eligibility, answering support — runs as software here, not headcount. That's how a small team operates a real-money-adjacent platform, and why the same engine can scale to a publisher's title or a media platform's broadcast without scaling people.
A platform that runs itself.
Twelve autonomous agents run most of what happens after a fan taps "predict" — not a back office. An event dispatcher routes every signal to the right specialist, and each one does a job a competitor staffs with a team:
12 live in production — more in development.
dispatcher
An AI resolver reads the gameplay outcome off the video and settles the pool — or refunds it. Winners paid by stake × skill, on an append-only ledger.
Disputes get an AI ruling with confidence-band routing. Clear cases resolve instantly; the gray-area ones escalate to a human.
Every risk flag is scored and actioned automatically — hold, freeze, or clear — and the system learns new fraud patterns every week.
Eligibility is re-verified on every pool and payout event. Anything that doesn't clear is held before money moves.
Tier-1 support tickets resolve in seconds, and a health agent watches the infrastructure around the clock.
Creators get proactive, AI-written coaching on what's working — so every contest runs a little better than the last.
Every ruling is logged on an append-only audit trail, and any decision the agents aren't confident about escalates to a human. It's leverage with a paper trail — see how the engine works.
One platform, four sides of the same loop.
A native revenue stream that turns viewers into participants. One tap launches a contest mid-stream — then the agent layer handles settlement, fraud checks, disputes, and support, so you just play.
Game knowledge becomes an edge. Predict outcomes, build an accuracy record, and earn a skill-weighted share of the pool — settled from the video, or refunded, with automated fraud and dispute checks.
Add skill-based engagement and monetization to your titles without building the contest engine, payout rails, AI operations, or compliance stack yourself.
Embed skill-based interactivity into live content to lift watch time and convert lurkers into active audiences.
Predictions are the first product — not the whole platform.
Skill-weighted predictions are live today and prove the model end-to-end: real engagement, credit payouts, real compliance — run by the agent layer above. Everything after is the platform expanding outward from that foundation.
Streamers run prediction contests during gameplay; fans' payouts are weighted by their accuracy. Settlement, fraud, disputes, compliance, and support all run as autonomous agents.
Agents that watch the live stream, detect prediction-worthy moments, and draft the next prompt — so a creator launches the right contest with one tap.
The same overlay, agent stack, and platform API that runs PLLAY today, offered to publishers and media platforms as embeddable, white-label infrastructure.
PLLAY is a skill-weighted prediction contest. We operate a skill-based contest framework with identity verification, geo-blocking, and age gating. 18+, eligible US states only. How skill works · Play responsibly
One round, end to end — settled from the broadcast itself.
It starts with a question.
A creator opens a prediction mid-stream. The pool starts at zero; fans pour in to call the next play.
Sharper reads earn more.
Every fan's slice is weighted by a 0.90×–1.40× skill multiplier from their track record. Skill, not luck — that's the whole game.
The round locks.
Calls close the instant the play begins. No late money, no edge from lag — everyone's committed at the same whistle.
It resolves from the video.
The outcome hits and the pool settles atomically — winners paid by stake × skill. No tickets, no invoicing, nobody scoring it by hand.
Everyone's paid — including you.
Winners take 80% of the pool. The creator keeps 20%. It's revenue that arrives the moment the round ends.
PLLAY is built and operated by PLLAY Labs, Inc., a Delaware corporation, founded by Shawn Gunn (CEO). Contests are sponsored by PLLAY Contests LLC, a wholly-owned subsidiary.
Reach us: support@pllay.io · @pllayme · security & trust · live status
Build with us.
Whether you create, watch, publish, or distribute live gameplay — there's a way to plug in.
Investor? See the thesis →