决策模型 Jev 上线一周，接入 Treg 客户线索筛选流程
Treg 找线索，Jev 负责打分筛选，比 Clay 便宜 85%，适合想搭销售线索智能体的人看看这套分工。
Jev 是一款决策模型，定位在智能体检索环节之后，对候选对象进行评分。在 Treg 的场景中，Treg 跨 60+ 人才数据供应商检索线索，Jev 再按用户设定的标准打分并输出结构化概率，让智能体自动判断哪些线索值得保留。
官方称其价格比 Clay 便宜 85%，并在 people search bench 准确率上排名第一。它提供插件形式，可接入任意智能体。
Jev is barely a week old, and this is already a pretty clean example of where a decision model fits inside an agent.
Clay for AI agents: you describe the kind of person you want, and @treg_ai searches across 60+ people-data providers to find matching leads and their details.
Jev then scores those people against your criteria, so the agent can automatically decide which leads are actually worth keeping. i.e. Treg handles the retrieval side > Jev sits after retrieval.
- 85% cheaper than Clay - #1 on people search bench accuracy - Plugin to any agent
It takes the candidate state and evaluates questions such as whether someone matches a role, company profile, or other criteria, returning structured probabilities rather than another paragraph of generated text.
本条情报归属 AITOP「产品」核心栏目，重点聚焦北美前沿实验室与硅谷科技巨头。在跨区域产业观察坐标中，该动态直接关联「AI 开发者工具链、Coding Agent 与工程化实践」研究主线，反映出当前技术演进与商业落地的最新风向。涉及核心实体与技术议题包括：Jev、Treg、Clay、智能体，读者可通过下方标签进入对应专题页查阅历史脉络与关联对比。
信源核验说明：本文原始线索采编自「rohanpaul_ai」（@rohanpaul_ai），经 AITOP 多维评分管线结构化提炼。
