Working, waiting, finished, or failed, grounded in reported activity.
See It In Action
Open Video ↗PlayLattice In Two Minutes
Discovery, local analysis, the Naive crew, player clips, TikTok, and AgentMail Outreach.
Recorded product walkthrough · 2:00 · On-screen captions · Original music. Source credits appear in the video. Recorded counts and service states reflect capture time.
Every shot.
A reason to watch.
A basketball content engine that turns shot clips into “make or miss?” moments. Give viewers a reason to guess, stay for the reveal, and share their take.
- 01 / Find
- Shot Moments Select the basketball action
- 02 / Ask
- Make Or Miss? Invite viewers to predict
- 03 / Hold
- Guess Before The Reveal Answer cards & a countdown
- 04 / Reveal
- Replay The Outcome Designed to spark conversation
Clip The ShotAsk The QuestionLet Them GuessRevealWorkflow Preview
Find The Shot Worth A Guess
The film room behind the content: select a moment and keep the source evidence.
From Basketball Footage To Audience Participation
Clip → Question → Prediction → Reveal| In The App | What You Can Do | Evidence |
|---|---|---|
| Playback & Tracking | Find candidate shot moments with player, ball, and hoop observations. | Model Predictions |
| Player Review | Inspect fragments, group tracks, and correct identity swaps. | Reviewer Confirmed |
| Highlights | Cut a short shot sequence while keeping the setup and outcome connected to the source. | Source Linked |
| Reviewed Stats | Confirm actor, event time, and outcome before adding totals. | Reviewed Events Only |
The Crew Behind The Content
Role overview · this static page does not show live agent status
| Agent | Responsibility | Handoff |
|---|---|---|
| CEO / Coordinator | Delegates work and assembles the package. | Specialist tasks → reviewed package |
| Scout | Checks source evidence and suitability. | Source context → clipper |
| Clipper | Selects moments and proposes quiz and music choices. | Selected clips → caption editor |
| Caption Editor | Prepares captions grounded in available evidence. | Caption package → review |
| Analyst | Checks package consistency and uncertainty. | Review notes → coordinator |
| Channel Manager | Prepares the distribution handoff. | Prepared package → local publisher |
What The Live Dashboard Shows
Tool calls, results, task messages, and agent handoffs.
Inspect outcomes, clear warning banners, and request cancellation.
Some crew runs use metadata only. A completed workflow does not establish visual verification. Distribution depends on an enabled policy and a working social provider.
Turn A Shot Into A Question
The engagement loop: pick an answer, wait for the reveal, then watch what happened.
PLAYLATTICE FILM QUIZ
Sample Question
Make or miss? Call it before the reveal.
Illustrative example · choose your prediction.
Example RevealIt Drops.
A real quiz would replay the source shot here. This is a format illustration, not a verified game result.
Build The Moment.
Invite The Guess.
- Clip A Basketball Shot
Keep enough of the setup to make the prediction interesting. Save the outcome for the reveal.
- Make Viewers Commit
“Make or miss?” Answer cards and a countdown invite a prediction before the play finishes.
- Reveal & Start A Conversation
Replay the outcome. Invite viewers to share their guess, compare reactions, or watch again.
This is the intended content format, not a claim of measured engagement lift. This interactive example uses an illustration; the demo above shows actual footage and the portrait preview.
Built Around A Local Core
Sports First| Layer | Implementation | Purpose |
|---|---|---|
| Film Room | React · TypeScript · Vite | Playback, tables, review, and live updates |
| API & State | FastAPI · SQLite | Persisted jobs, review records, and media access |
| Video Models | YOLO26 pose · E-BARD RF-DETR | Player, pose, basketball, and hoop observations |
| Exports | FFmpeg | Source-linked clips and portrait renders |
| Agent Workflow | Naive | Specialist sessions, handoffs, and structured outputs |
Current Boundaries
Ball and hoop detections can be sparse. Player tracks can fragment or swap. Highlight suggestions require review; they do not establish official statistics or verified athlete identities.
The prototype has been exercised on short clips, not validated as full-game coverage. Crew output can be metadata-only. This project page runs independently of the analysis app.
Explore The Architecture ↗ Setup Guide ↗Call The Shot On TikTok
See the clips. Make your prediction. Join the conversation.
Watch On TikTok ↗
Recorded Post Preview
The Content, Without The Embed.
An actual portrait quiz published by @playlatticeai. This recorded GIF keeps the example visible even when TikTok’s player doesn’t load.
This earlier format asks viewers to predict player movement. The make-or-miss interaction above illustrates the shot-quiz direction.
Open The Original Post ↗Footage credit: Singpore amatuer basketball channel · Original Source. Excerpted, cropped, and overlaid with quiz graphics.
Show TikTok Profile Embed
If the profile doesn’t load, open it on TikTok.
PlayLattice
More than a highlight.
A reason to participate.
Built for the YC Naive Hackathon. Sports First.
Explore The Project ↗Built with
