Claude Haiku 3
Anthropic · claude-3-haiku · 有条件适配
人工参与：高风险、低置信度和投诉场景由人工坐席确认或接管。该关联表示公开案例中观察到的使用，不代表通用推荐或独立效果评估。来源：AppFolio and Claude Platform ↗对话摘要、知识检索、回复建议、质检和工单分类。
年鉴当前收录 3 个有公开使用证据的模型、8 个企业案例和 20 项数值指标。
这些关联表示公开案例中的实际使用或条件性适配，不是无条件推荐；人工参与要求为：高风险、低置信度和投诉场景由人工坐席确认或接管。
ROI 只有在投入、持续成本、收益、周期和计算方法同时公开时才展示；当前缺项不会由年鉴自行估算。
依次确认工作流与人工边界、来源支持的模型关系，再审视企业案例和指标；这不是模型推荐、上线批准或 ROI 承诺。
Anthropic · claude-3-haiku · 有条件适配
人工参与：高风险、低置信度和投诉场景由人工坐席确认或接管。该关联表示公开案例中观察到的使用，不代表通用推荐或独立效果评估。来源：AppFolio and Claude Platform ↗Anthropic · claude-3-haiku · 有条件适配
人工参与：高风险、低置信度和投诉场景由人工坐席确认或接管。该关联表示公开案例中观察到的使用，不代表通用推荐或独立效果评估。来源：Assembled Claude Platform (API) case study ↗Anthropic · claude-3-5-sonnet · 有条件适配
人工参与：高风险、低置信度和投诉场景由人工坐席确认或接管。该关联表示公开案例中观察到的使用，不代表通用推荐或独立效果评估。来源：AppFolio and Claude Platform ↗Anthropic · claude-3-5-sonnet · 有条件适配
人工参与：高风险、低置信度和投诉场景由人工坐席确认或接管。该关联表示公开案例中观察到的使用，不代表通用推荐或独立效果评估。来源：Assembled Claude Platform (API) case study ↗Google DeepMind · Gemini 1.5 Pro · 有条件适配
人工参与：高风险、低置信度和投诉场景由人工坐席确认或接管。该关联表示公开案例中观察到的使用，不代表通用推荐或独立效果评估。来源：Gelato: Bringing efficient, high-quality local production to global ecommerce with Gemini ↗业务场景：客服坐席辅助
AppFolio identifies Claude 3 Haiku as the high-volume, rapid-response model in Realm-X Messages. The vendor case reports almost tripled adoption after the Haiku and infrastructure changes, about 11 hours saved per property manager each week, more than half a million AI-suggested responses processed, and an average 26-second reduction in message response time.
ROI：公开投入、持续成本、收益或测量周期不足，未计算。
边界：Vendor-authored case; the reported outcomes describe a two-model platform and infrastructure revamp. No sample definition, measurement period, independent validation or model-only causal estimate is published. The deployment stage is kept unknown because the page does not explicitly classify it as production or scaled.
业务场景：客服坐席辅助
AppFolio identifies Claude 3.5 Sonnet as the model for complex communications that incorporate resident history, maintenance tickets, property-specific policies and amenity information.
ROI：公开投入、持续成本、收益或测量周期不足，未计算。
边界：Vendor-authored case; the page does not publish Sonnet-specific volume, latency, quality, cost or ROI measurements. The quantitative results are retained only on the shared platform/Haiku view and are not attributed to Sonnet 3.5.
业务场景：客服坐席辅助
ASAPP says its GenerativeAgent achieved a 25-40% improvement in core business outcomes compared with other AI models in production after switching to Claude in Amazon Bedrock; the exact Claude version is not published.
ROI：公开投入、持续成本、收益或测量周期不足，未计算。
边界：Vendor-authored case; no specific model/version can be bound, so this case remains organization-level. The reported range is not converted into an efficiency percentage or ROI figure because the underlying metric definitions and methods are not disclosed.
业务场景：客服坐席辅助
Assembled explicitly identifies Claude 3 Haiku as the high-speed operations model in Assembled Assist. The case reports platform and customer support outcomes including higher CSAT, fewer escalations, more cases solved per hour, higher case automation, and customer examples from Thrasio and Honeylove.
ROI：公开投入、持续成本、收益或测量周期不足，未计算。
边界：Vendor-authored case; Assembled Assist uses both Claude 3 Haiku and Claude 3.5 Sonnet and routes work by complexity. Metrics are attached to the shared platform view for discoverability, not presented as isolated Haiku causal estimates. Customer names and results are not independently audited on the page.
业务场景：客服坐席辅助
Assembled explicitly identifies Claude 3.5 Sonnet as the complex-reasoning model in Assembled Assist, while Claude 3 Haiku handles high-speed operations and the platform routes cases to the appropriate handler.
ROI：公开投入、持续成本、收益或测量周期不足，未计算。
边界：Vendor-authored case; no Sonnet-specific volume, latency, quality, cost or ROI measurement is published. Shared platform and customer metrics are not duplicated into this Sonnet view.
业务场景：客服坐席辅助
Mistral says AI use cases assist cargo release, fact-checking, client information requests and shipping-document processing, targeting faster and higher-quality responses and improved customer-facing processes.
ROI：公开投入、持续成本、收益或测量周期不足，未计算。
边界：Vendor-authored case; no named model/version, deployment stage, baseline, measurement method, numeric result or independent validation is published.
业务场景：客服坐席辅助
Google Cloud's public case names Gemini 1.5 Pro in Gelato's engineering-support ticket triage and customer-support error categorization. It reports 120 hours of weekly manual tasks saved and ticket-assignment accuracy increasing from 60% to 90%; Gemini automatically assigns tickets to the responsible engineering team. Gelato also describes instant error classification in customer-support systems.
ROI：公开投入、持续成本、收益或测量周期不足，未计算。
边界：Vendor-authored Google Cloud case; the reported 120 hours, accuracy figures and support-system result are not isolated from Vertex AI, data preparation, workflow integration or human review. The page does not publish a dated API model ID, sample, evaluation method, measurement period, cost, quality beyond assignment accuracy or independent validation. The separate plan to use Gemini 2.0 for image-based product quality is not bound to this Gemini 1.5 Pro case.
业务场景：客服坐席辅助
Hershey Entertainment & Resorts uses Google Cloud Customer Engagement Suite, Conversational Agents, Vertex AI, Gemini and connected enterprise systems to automate guest and employee support. The public case reports chatbot launch in less than 40 hours, more than 25% of employee call-off requests automated, 50%+ of password-reset calls managed, and eight hours of employee work saved per day.
ROI：公开投入、持续成本、收益或测量周期不足，未计算。
边界：Vendor-authored Google Cloud case; no Gemini model/version is named, so this case remains organization-level. The contribution of Gemini versus Dialogflow/Conversational Agents, BigQuery, webhooks and workflow integrations is not isolated. Sample, period, cost, quality, escalation and independent validation are not disclosed.