AI 投放合作 · 海外 & 国内AI-powered growth · Global & China

增长团队的 AI 化改造
一人 + AI,发挥一整个团队的效能。
AI transformation for growth teams.
One person + AI. The capability of a team.

面向 AI 产品与增长团队,提供从投放流程梳理、媒体 API 接入到 AI 工作流落地的改造服务。让 AI 承接数据分析、策略建议、投放执行与效果复盘,帮助一个人协同 AI 完成过去需要多人配合的增长工作。Help AI products and growth teams adopt AI across their advertising operations, from mapping processes and connecting media APIs to building working AI workflows. With AI handling analysis, strategy recommendations, campaign execution and performance reviews, one person can take on growth work that once required a team.

关于我About me

我有 6 年海外用户增长与 PC 增长经验,曾在多家大型团队从 0 到 1 负责 PC 增长、归因系统和投放链路,也负责过豆包、Coze 等 AI 产品的渠道增长。现在以 FDE 的方式推进投放 AI 化:连接数据与工具,把业务经验写成可执行规则,并在团队内开展 AI 工作流培训。I bring six years of experience in overseas user acquisition and PC growth, building PC acquisition, attribution systems and conversion journeys from the ground up across major teams. My work also includes channel growth for AI products such as Doubao and Coze. Today, I deliver AI adoption through FDE: connecting data and tools, turning operating experience into executable rules, and training teams to use AI workflows.

$1B+
累计操盘媒体消耗Cumulative media spend managed
AI 工具 × 游戏AI × Gaming
豆包 / Coze · 大型游戏项目Doubao / Coze · Major gaming projects
海外 & 国内Global & China
跨市场获客与投放优化Acquisition and optimization across markets
阅读投放与 AI 工作流实践 · FDE ↗Explore growth workflows · FDE ↗

AI 投放优化:从数据分析到自动执行。AI ad optimization, from analysis to execution.

把每天重复的数据处理、投放判断和账户操作,连接成一套可执行、可追踪、可复盘的 AI 工作流。Connect recurring data work, campaign decisions and account operations in an AI workflow you can run, trace and review.

01

数据接入与自动分析Connect and analyze

以 Claude Code、Codex CLI 为工作入口,连接 Google、Meta、TikTok、Yandex 等媒体 API、BI 数据与飞书。打通取数、分析、日报和异常告警。Use Claude Code and Codex CLI to connect ad-platform APIs, including Google, Meta, TikTok and Yandex, with BI data and Feishu. Build a working flow for reporting, analysis and alerts.

02

AI 策略判断Turn data into decisions

回看历史人工操作,按媒体、广告类型和生命周期整理适用条件。将目标、判断信号、操作范围和复盘标准写入分层 Skills 与规则文件。Review historical operator decisions by platform, campaign type and lifecycle stage. Encode objectives, decision signals, permitted actions and review criteria in layered skills and rule files.

03

自动执行与持续复盘Execute and improve

操作前记录预测与证伪条件,执行后读回核验,并按数据成熟度安排复盘。将这套工作流扩展到网页 SEO、落地页巡检与 Seedance 短视频生产。Record expectations and failure conditions before acting, verify changes after execution, and review outcomes when the data is ready. Extend the workflow to SEO, landing-page checks and short-form video production with Seedance.

全自动运行,按业务设定边界。Automation shaped around your business.预先约定目标、预算和操作范围,重要变更可保留人工审核。Define goals, budgets and operating scope up front, with human review for significant changes where needed.

FDE 落地交付,让 AI 接入增长业务。FDE delivery: bring AI into your growth operations.

从真实投放问题出发,先搭建数据与工具基建,再设定业务规则,最后用持续复盘验证规则。交付的是团队能长期运行、检查和迭代的增长工作流。Start with the operating problem, connect the data and tools, define the rules, then evaluate them through ongoing reviews. The deliverable is a growth workflow the team can operate, inspect and improve.

01 / INFRASTRUCTURE

基建:接通数据与工具Connect data and tools

打通媒体 API、BI、飞书与素材工具,明确数据口径、账户权限和执行回执,让取数到执行的链路可靠运行。Connect media APIs, BI, Feishu and creative tools. Define metrics, account permissions and execution checks so the full operating flow works reliably.

02 / RULE SETTING

规则 Setting:沉淀操作逻辑Set the operating rules

按媒体 × 广告类型 × 生命周期归纳历史操作,定义信号、动作、预测与证伪条件,将经验转成可调用的规则。Review past actions by platform, campaign type and lifecycle. Define signals, actions, expectations and failure conditions as reusable operating rules.

03 / ITERATE

量化式复盘:检验与迭代Evaluate and iterate

参考量化研究的评估思路,持续记录预测与结果,区分判断、执行和环境变化,在可比样本上验证后再扩大自动化范围。Borrow the evaluation discipline of quantitative research: track predictions and outcomes, separate decision and execution errors from changing conditions, and expand automation after validation on comparable samples.

阅读 FDE 化思路:从基建到规则 Setting ↗Read the FDE approach: infrastructure and rule setting ↗