jev-buzzword-rush
25-second Japanese IT-buzzword game that shows Jev's speed and cost: every typed term is classified live, with no LLM-generated text.
Overview
Jev Buzzword Rush is a small Japanese game (with an English section) designed to let people feel how fast and cheap Jev is. You type or speak IT terms with no submit button, and Jev classifies each into a field (cloud, networking, databases, programming, security, or not an IT term); two terms per field fill its meter and filling all five clears the game. The screen shows the running cost on one side and the actual HTTP requests and responses on the other.
In the same parallel request Jev also judges how current a term is, how costly it is to learn, how niche it is, and which trivia item it matches, and code picks a prepared reply from those judgments; no text is generated by any model. Code handles candidate extraction, scoring, reply selection and repetition detection. The README reports one demo run with 14 judgments averaging about 400 ms and a total cost of about 0.14 yen. It follows TypeSafe's guidance to select rather than generate, compute in code and batch independent questions. It runs with Python 3.10+ standard library and a local relay server holding the key, and is MIT licensed.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which IT field does this term belong to? | cloud · network · database · programming · security · not an IT term |
| Score | How current / how costly to learn / how niche is this term? | — |
| Choice | Which prepared trivia item does this term match? | — |
How it works
- User types or speaks a term
- Code extracts candidates and checks repeats
- One parallel Jev request classifies and rates the term
- Code updates meters and selects a prepared reply
- Cost and raw requests shown live
Strengths
- Makes latency and per-call cost tangible
- No generated text; replies are selected
- Key kept on a local relay server
Caveats
- A demo game
- Japanese-language content
Getting started
`TYPESAFE_AI_KEY=... python3 server.py`, then open http://localhost:8787.At a glance
- Type
- Project
- Status
- prototype
- Language
- Python, JavaScript
- Runtime
- local server and browser
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
A 25-second IT buzzword game to feel how fast and cheap Jev (TypeSafe System One) is — code + Jev only, no LLM.
logicrw/awesome-jev-projects
Frequently asked questions
What does Jev decide?
Which IT field a term belongs to, plus ratings and trivia matches used to pick replies.
Does any model write the replies?
No, code selects prepared replies based on Jev's judgments.
How much does a game cost?
About 0.14 yen for a demo run, per the README.
Related projects
Public Jev lab centred on a chess match broadcast where Jev picks from legal moves, plus smaller judgment booths and a Fast Flag catalogue.
ProjectprototypeJavaScript, PythonVisual workbench (in Portuguese) for wiring Jev decision funnels as node graphs and watching them play a game in real time.
ProjectprototypePythonTic-tac-toe teaching project comparing Jev Choice, Score and Noul designs across three levels of state detail.
ProjectresearchJavaScriptJapanese quiz demo where Jev guesses which of AWS, Azure or Google Cloud a debranded service name belongs to, showing its probabilities.
DemoprototypeTypeScriptStreamlit Tetris demo where Jev picks one collision-checked placement ID and Python verifies cells before locking.
AppprototypePythonWeb lab where Jev plays 2048, showing each move's probability distribution, confidence, latency and token cost.
Demoprototype