SELECT AI_COMPLETE('ai_gateway_conn:moonshotai/kimi-k2.6',
'hello, introduce yourself in one sentence'
);
返回（实测 3.1 秒）：
I am Kimi, a large language model created by Moonshot AI, here to help you with information, writing, analysis, and conversation.
通用 Prompt 模板
-- 摘要生成
SELECT AI_COMPLETE('conn:model',
CONCAT('用一句话概括：', content)
) FROM articles;
-- 情感判断
SELECT AI_COMPLETE('conn:model',
CONCAT('判断情感（positive/negative/neutral），只回复一个词：', review)
) FROM reviews;
-- 内容分类
SELECT AI_COMPLETE('conn:model',
CONCAT('将以下文本归类为 sports/finance/tech/health：', text)
) FROM contents;
-- 翻译
SELECT AI_COMPLETE('conn:model',
CONCAT('将以下英文翻译为中文：', english_text)
) FROM docs;
💡 技巧：Prompt 中加"只回复一个词"或"只回复类别名称"可控制输出格式，方便后续 GROUP BY 或 WHERE 过滤。
第三步：实战 —— AI 增强 NHL 数据分析
场景 1：球员球探报告
将 Gold 层聚合数据（生涯进球、助攻、得分）输入 LLM，生成专业球探评价：
SELECT
player_name,
position,
total_goals,
total_points,
AI_COMPLETE('ai_gateway_conn:moonshotai/kimi-k2.6',
CONCAT('用一句话中文评价这位NHL球员的职业生涯风格：',
player_name,
'，位置', position,
'，生涯进球', CAST(total_goals AS STRING),
'，总得分', CAST(total_points AS STRING))
) AS ai_scout_report
FROM gold.player_career_stats
ORDER BY total_points DESC
LIMIT 5;
SELECT
team_abbr,
wins, losses, goals_for, goals_against,
AI_COMPLETE('ai_gateway_conn:moonshotai/kimi-k2.6',
CONCAT('用一句话中文总结这支NHL球队的赛季表现：',
team_abbr, '，赛季2019-20，',
'胜场', CAST(wins AS STRING),
'，负场', CAST(losses AS STRING),
'，进球', CAST(goals_for AS STRING),
'，失球', CAST(goals_against AS STRING))
) AS season_review
FROM gold.team_season_summary
WHERE season = 20192020
ORDER BY wins DESC
LIMIT 5;
实测结果（5 行，10.1 秒）：
球队
AI 赛季评述
TBL (闪电)
表现极为出色，进攻火力强大，防守稳健，最终夺得斯坦利杯总冠军
VGK (金骑士)
表现中规中矩，进攻略优于防守，整体胜率超过五成
COL (雪崩)
进攻火力尚可但防守不稳，整体胜率接近五成
场景 3：情感分析 + 分类
对任意文本做即时分析，无需预训练模型：
SELECT
AI_COMPLETE('ai_gateway_conn:moonshotai/kimi-k2.6',
'判断情感（positive/negative/neutral），只回复一个词：'
|| 'The Lightning played an incredible game tonight!'
) AS sentiment_1,
AI_COMPLETE('ai_gateway_conn:moonshotai/kimi-k2.6',
'判断情感（positive/negative/neutral），只回复一个词：'
|| 'Terrible officiating tonight.'
) AS sentiment_2,
AI_COMPLETE('ai_gateway_conn:moonshotai/kimi-k2.6',
'将文本归类为 sports/finance/tech/health。只回复类别：'
|| 'NHL playoffs start next week with 16 teams.'
) AS category;
实测结果：
sentiment_1
sentiment_2
category
positive
negative
sports
第四步：批量 AI 处理模式
模式 A：逐行处理
适合小数据集（< 1000 行），直接 SELECT + AI_COMPLETE：
SELECT id, content,
AI_COMPLETE('conn:model', CONCAT('总结：', content)) AS summary
FROM articles
WHERE created_date = CURRENT_DATE();
模式 B：Dynamic Table 自动批处理
将 AI 分析结果沉淀为 Dynamic Table，后续查询零延迟：
-- 创建 AI 增强分析表
CREATE OR REPLACE DYNAMIC TABLE gold.player_ai_scouting
REFRESH INTERVAL 1 DAY vcluster DEFAULT
COMMENT 'AI 球探报告——每日自动更新'
AS
SELECT
player_id,
player_name,
position,
total_goals,
total_points,
pts_per_game,
AI_COMPLETE('ai_gateway_conn:moonshotai/kimi-k2.6',
CONCAT('用一句话中文评价这位NHL球员：',
player_name, '，位置', position,
'，进球', CAST(total_goals AS STRING),
'，得分', CAST(total_points AS STRING),
'，场均得分', CAST(pts_per_game AS STRING))
) AS scouting_report
FROM gold.player_career_stats
WHERE total_points > 500; -- 只分析得分 500+ 的球员，控制成本
-- 应用层直接查询，零 AI 调用延迟
SELECT player_name, scouting_report
FROM gold.player_ai_scouting
WHERE player_name = 'Alex Ovechkin';
⚠️ 注意：DT 每次 REFRESH 会对所有符合 WHERE 条件的行重新调用 AI_COMPLETE。建议：
