推测性扇出

推测性扇出

在一次调用里发送许多问题（包括推测性的），由你的代码决定哪些真正相关。

因为 TypeSafe 支持在单次 API 调用里发送多个问题，我们建议把系统需要的所有问题都放进同一个请求，事后再用代码决定哪些相关。所有问题都是并行求值的，所以多问几个通常对响应时间没什么影响。

示例：支持工单分流

设想你在做一个支持系统，需要给支持工单分流。你要先把工单归到一个类别。如果是 bug 报告，还要判断 bug 的严重程度。

与其先问类别、再发一次调用问严重程度，你可以在同一次里把两个都问出来。如果工单不是 bug 报告，直接忽略 bug 严重程度那个问题的结果就行。

%%{init: {"fontFamily": "Inter, sans-serif", "flowchart": {"rankSpacing": 35, "wrappingWidth": 300, "subGraphTitleMargin": {"top": 8, "bottom": 60}}}}%%
flowchart LR
    t["support ticket"]

    subgraph req["TypeSafe AI model<br/>evaluates each question<br/>against the ticket in parallel"]
        direction TB
        c["<b>Choice:</b> category"]
        b["<b>Score:</b> bug severity"]
        r["<b>Noul:</b> reproducible steps?"]
        f["<b>Noul:</b> refund requested?"]
        s["<b>Score:</b> frustration"]
        %% invisible links: without an edge these share a rank and sit side by side
        c ~~~ b ~~~ r ~~~ f ~~~ s
    end

    t -- "one request<br/>ticket + 5 questions" --> req
    req -- "one response: 5 answers<br/>decisions + probabilities" --> route{"<b>filter, combine, and route</b><br/>in your code"}
    route -- "bug_report" --> eng["read severity + repro steps<br/>escalate or backlog"]
    route -- "billing" --> bill["refund requested<br/>send to billing"]
    route -- "feature_request" --> feat["log it<br/>sent to devs"]

第 1 步：推测性扇出

{
  "category": {
    "type": "choice",
    "instructions": "Determine the broad category of this support ticket",
    "criteria": {
      "bug_report": "The user is reporting something that is broken or producing errors",
      "billing": "Charges, invoices, refunds, subscriptions",
      "feature_request": "The user is requesting new functionality",
      "account": "Login, permissions, profile, security"
    }
  },
  "bug_severity": {
    "type": "score",
    "instructions": "How severe is the reported issue",
    "criteria": [
      "Cosmetic; no impact to functionality",
      "Broken or degraded feature; workaround exists",
      "Blocking issue; no workaround exists"
    ]
  },
  "has_reproducible_steps": {
    "type": "noul",
    "instructions": "The user describes specific steps to reproduce the issue"
  },
  "refund_requested": {
    "type": "noul",
    "instructions": "The user is explicitly asking for a refund or credit"
  },
  "frustration": {
    "type": "score",
    "instructions": "How frustrated the user appears",
    "criteria": [
      "Calm, matter-of-fact",
      "Frustrated but civil",
      "Very angry"
    ]
  }
}

第 2 步：用代码路由

你的代码根据分类结果决定哪些相关：

triage.py

category = response.answers["category"]
bug_severity = response.answers["bug_severity"]
bug_repro = response.answers["has_reproducible_steps"]
refund = response.answers["refund_requested"]
frustration = response.answers["frustration"]

if category.choice == "bug_report":
    if bug_severity.score > 1.5 and bug_repro.noul > 0.6:
        escalate_to_engineering(ticket_id, severity="high")
    else:
        add_to_bug_backlog(ticket_id)

elif category.choice == "billing":
    if refund.noul > 0.7:
        route_to_billing_with_flag(ticket_id, refund_likely=True)
    else:
        route_to_billing(ticket_id)

elif category.choice == "feature_request":
    log_feature_request(ticket_id)

# Frustration is useful regardless of category
if frustration.score > 1.5:
    flag_for_priority_response(ticket_id)

整棵决策树所需的一切都来自一次调用。推测性问题在不相关时被忽略，在相关时则省下了一次往返。