应用产品
通过语音和指点在 tldraw 画布上绘画,Jev(TypeSafe System One)每个词约 350 毫秒内确定动作、目标和位置。
Draw on a tldraw canvas with your voice and a pointing finger. Jev (TypeSafe System One) decides action, target and place in ~350 ms per spoken word.
<Card title="JEV" note={view.busy ? 'asking…' : jev ? `${jev.latencyMs} ms` : ''} tone={jev?.verdict}>链接指向我们抓取当天的那个 commit,行号是准的。
下面是从这个项目源码里原样取出来的 question 组合。
is_command是/否Is the speaker giving an instruction to the drawing canvas? The speaker may use English or Ukrainian.
complete是/否Has the speaker finished the instruction, so that it can be carried out right now?
import { experimental_evaluate as evaluate } from 'ai';
const { answers } = await evaluate({
model: 'typesafe-ai/jev',
state,
questions: {
is_command: { type: 'boolean', instructions: 'Is the speaker giving an instruction to the drawing canvas? The speaker may use English or Ukrainian.' },
complete: { type: 'boolean', instructions: 'Has the speaker finished the instruction, so that it can be carried out right now?' },
},
});