模型每代都在漲。落地卻沒動。 Models keep getting smarter. Landing doesn't move.
Klarna 砸了 4000 萬美元給 OpenAI 做 AI 客服，一年後退回人機混合。麥當勞跟 IBM 做 AI 點單，三年後撤掉。中間那條鴻溝，模型公司不解決，諮詢公司沒工具。 Klarna spent $40M on OpenAI's customer-service AI. One year later, they walked it back. McDonald's pulled IBM's AI ordering after three years. Model companies stop at the API. Consultancies don't have the tools.
我們用 AI 實施科學（AI Implementation Science）——醫學界 20 年的方法論，今天用在 AI 上。 We bring AI Implementation Science to the enterprise—a 20-year discipline from medicine, now applied to AI.
剩下 80% 是看不見的成本——流程、數據、組織、監管、維護。
The other 80% is invisible—process, data, organization, regulation, maintenance.
技術門檻降低了，認知門檻凸顯了。把專家的隱性知識結構化教給 AI——這個職業的時間窗口只有 2-3 年。
Technical barriers fall. Cognitive barriers rise. Structuring tacit expertise into something AI can carry has a 2-3 year window.
做 AI 的人有武器，沒戰場。做傳統行業的人有戰場，卻在等死。機會在懂 AI 又懂行業的人手裏。
AI builders have weapons but no battlefield. Industry incumbents have a battlefield but are dying on it. The opportunity belongs to those who understand both.
未來的醫療 AI，不在矽谷的數據中心裏。在你家樓下診所的那台伺服器上。
The future of medical AI isn't in a Silicon Valley data center. It's on the server in your neighborhood clinic.
10 分放大到 50 分。0 分放大完還是 0。差距不在工具，在元認知。
10 amplified is 50. 0 amplified is still 0. The gap isn't in the tool—it's in metacognition.
CFIR 改造版——醫學界 20 年的診斷框架，加上 AI 特有的三個參數：迭代速率、監管漂移頻率、角色重疊度。診斷完才決定怎麼幹。上線後用 RE-AIM 維護。 A CFIR adaptation—twenty years of medical implementation framework, with three AI-specific parameters: iteration velocity, regulatory drift, role overlap. Diagnose first. Decide what to build after. Maintain with RE-AIM post-launch.
agent 能力邊界、成本曲線、迭代速率
Agent capability, cost curve, iteration velocity
監管漂移頻率、行業壓力、合規邊界
Regulatory drift, industry pressure, compliance limits
數據資產、流程兼容性、工具棧現狀
Data assets, workflow compatibility, tooling
角色重疊度、心理動力、團隊學習曲線
Role overlap, motivation, team learning curve
試點 → 灰度 → 全量 → 維護
Pilot → gradual → full → maintenance
五維診斷 + 分階段路線圖 + 試點立項書。4-8 周交付。
Five-Dimension Diagnostic + phased roadmap + pilot charter. 4-8 weeks.
瞭解更多 → Learn more →1-6 個月全程支持。深入企業一線，每週對齊。RE-AIM 框架管落地後的持續調優。
1-6 months of full-stack support. On-site weekly. RE-AIM framework manages post-launch evolution.
瞭解更多 → Learn more →工程團隊親自講。線上 / 線下 / 內訓 / 公開課。可定製行業版本。
Taught by our engineering team. Online / on-site / internal / open. Industry-customized.
瞭解更多 → Learn more →業務場景定製 + 系統集成（IM / OA / CRM / 知識庫）+ 上線運維。
Custom-fit + system integration (IM / OA / CRM / KB) + production ops.
瞭解更多 → Learn more →合規、隱私、本地部署是硬約束。我們是醫學博士牽頭的團隊。
Compliance, privacy, local deployment are hard constraints. We're led by a medical PhD.
人才培養與課程體系才是壁壘。教育線由國家級優秀指導教師與萬人學習社羣運營者牽頭。
Talent and curriculum are the moat. Led by a National Outstanding Instructor and a 10,000+ community operator.
200 國 11,000+ 銀行金融平台是我們團隊跑過的活。
Financial platforms for 11,000+ banks across 200 countries.
34 個站點廣告數據分析、AI 投流優化、多語言內容質控全套都做過。
Ad analytics across 34 sites, AI bid optimization, multilingual content QC at scale.
字節億級 DAU 應用 + Fortune 500 製造場景跑過。Industrial AI 與算力調優一體交付。
ByteDance 1B+ DAU systems + Fortune 500 manufacturing experience. Industrial AI and compute optimization in one stack.
簡歷初篩與匹配打分自動化。
Resume screening and candidate matching automated.
新媒體運營 Agent 接管選題與文案。
Social media agent took over topic and copy.
3 個月陪跑，從場景篩選到 Agent 部署。
3-month engagement, from scenario selection to agent deployment.
Fortune 500 / 200 國 11,000+ 銀行 / 字節億級 DAU 應用 / 高校 RAG 文獻平台
Fortune 500 / 11,000+ banks across 200 countries / ByteDance 1B+ DAU apps / academic RAG platforms
國家級優秀指導教師 / ICPC 銀牌 / 已授權專利 / 阿里高級研發 / 字節資深架構
National Outstanding Instructor / ICPC Silver / granted patents / former Alibaba Senior R&D / former ByteDance Senior Architect
AI 實施科學提出者 · 昆士蘭大學醫學博士
Proposer of AI Implementation Science · Medical PhD, University of Queensland
[ 待 Roland 補：閲讀總量 / 引用次數 / 媒體轉載 ]
[ TBD: total readership / citations / press mentions ]
[ 待 Roland 補：已合作客戶 logo 矩陣 ]
[ TBD: client logo grid ]
Anthropic / OpenAI / Claude Code / 國產模型矩陣
Anthropic / OpenAI / Claude Code / domestic model fleet
過去兩年，AI 行業在問「模型能做什麼」。下一個五年，真正的問題只有一個：你能讓它被做成嗎。Roland 押注未來五年職業方向的核心論點。CFIR / NPT / RE-AIM 等醫學實施科學框架系統性翻譯到 AI 場景。 For two years, the AI industry asked: what can the model do? For the next five, only one question matters: can you get it done? Roland's five-year career bet, articulated. A systematic translation of CFIR / NPT / RE-AIM from medicine to AI.
下載完整 PDF →Download full PDF →AI 實施科學提出者。「這是我賭上接下來五年職業方向的一件事。」 Proposer of AI Implementation Science. "This is the bet I'm placing on my next five years."
工程團隊曾在阿里寫訂單交易系統、在字節做億級 DAU 應用架構、為 CapCut 與 Dreamina 設計生成式系統、在 Fortune 500 交付過億元 AI 項目、為全球 200 國 11,000+ 銀行做金融平台、為高校做文獻平台。
Our engineering team has built order systems at Alibaba, billion-DAU architecture at ByteDance, generative systems at CapCut and Dreamina, hundred-million-RMB AI projects across Fortune 500 clients, financial platforms for 11,000+ banks across 200 countries, and academic RAG platforms.
完整陣容 →Full team →