Everyone Has “1,000 Employees”

After the release of AI Leadership, we hosted two live-streamed discussions, one on Liu Run’s channel and another on WaytoAGI, receiving strong positive feedback.

Two years ago, when Xiang Yang and I began writing this book, AI had fewer capabilities than it has today. We already sensed something fundamental: to use AI well and truly lead it, you must first change how you see it.

If you treat AI as a digital employee and collaborative partner, you naturally begin setting goals, breaking down tasks, defining standards, conducting reviews, and orchestrating multiple agents to tackle complex work together.

That perspective became the book’s central thread.

We frame AI leadership as a 3M model: Method × Mindset × Maturity

Methods include prompting, knowledge bases, and iterative task design;
Mindsets encompass logic, abstraction, and creativity;
Maturity grows from experience, aesthetic judgment, contextual discernment, and persistent questioning.

Section 1.2, titled “Everyone Has ‘1,000 Employees’,” centers on the “1+N” working model: one person sets direction, defines standards, and bears final accountability, while multiple agents run parallel research, execution, validation, and iteration. We introduced this framework two years ago. Today, it is becoming mainstream practice.

That’s something we’re proud of.

When writing the book, we held ourselves to two standards:

Five years later, its core arguments should still hold up, logically sound and practically relevant.
Readers should be able to start applying ideas immediately and turn the theory into changed behavior.

Tools and models will keep evolving. What endures, and what ultimately determines the quality of human-AI collaboration, is human expression, judgment, experience, and creative insight.

If this book helps you begin with one real task, shift AI from an occasional tool to a trusted co-worker, gradually build your own 1+N workflow, clarify your unique role and responsibility within it, understand where your agency ends and AI’s begins, and learn to orchestrate multiple AIs to construct complex outputs and systems, then our work has fulfilled its purpose.

You can order AI Leadership on JD.com.

The “1+N” Working Model

Xu Xin of Today Capital recently said in a podcast: “In the future, many organizations may have no middle management at all. A new structure will emerge, with one highly capable individual directing AI agents and dramatically boosting efficiency.”

That aligns closely with the foundational logic in AI Leadership.

The “1+N” model, 1 human + N AI agents, is fast becoming the default mode of knowledge work.

Three key implications follow:

Reframe AI as a colleague: Treat AI as a digital employee capable of collaboration, execution, and delivery.
Rethink task scope: With that mindset, you’ll intentionally design workflows where one person guides multiple agents to accomplish tasks once requiring full teams, and as models and tool ecosystems mature, those tasks will grow steadily more sophisticated.
Clarify roles: In the 1+N model, the human owns vision, standards, and accountability; agents handle parallel execution across research, drafting, verification, and iteration.

When Xu Xin says “a highly capable individual,” she means someone skilled in goal-setting, task decomposition, context structuring, standard definition, outcome evaluation, and ownership, even when things go wrong.

As AI absorbs more execution, these human capabilities become more scarce. The 1+N model also multiplies leadership leverage: one person’s judgment now governs dozens of concurrent, high-fidelity operations.

Transparent Sales Management in Practice

We recently hosted the CEO of a data analytics firm.

His company turned around dramatically, from losing several million RMB annually to generating several million in profit each year.

He shared rich, actionable insights on operational discipline and team management.

One practice stood out: his transparent, incentive-aligned sales management system.

Every salesperson has their own dashboard, showing personal KPIs and anonymized peer metrics (e.g., conversion rate, average deal size, cycle time).

Salespeople define their own quarterly targets, and base salary scales directly with those targets. Performance bonuses and commissions are fully public, calculated by clear, pre-agreed formulas. Even individual income figures appear on the dashboard, in real time.

Of course, transparency alone isn’t enough. He built guardrails:

Target ranges are bounded (min/max thresholds per role/seniority);
Quarterly performance reviews trigger automatic base-salary adjustments, if targets aren’t met, base pay resets downward;
Managers review self-set targets for realism: new hires, for example, can’t claim top-tier quotas without historical performance to justify them, preventing short-term “target arbitrage”;
Progress is tracked weekly; underperformance triggers coaching, then structured exit paths;
Lead distribution follows explicit rules, weighted across lead source, recency, fit score, and historical conversion, while outcomes are measured across multiple dimensions: revenue, gross margin, customer quality coefficient, and collections timeliness.

The biggest win was eliminating nearly all negotiation overhead around goals and compensation. Each salesperson effectively runs a micro-profit center with full P&L visibility, autonomy, and accountability. It’s a compelling case of internal marketization done right.

A Sales Story

This GEO sales example is revealing.

A business leader from a traditional enterprise was evaluating GEO services.

During discovery, our sales rep mentioned “RAG” and “vectorization.” The prospect asked: “What is vectorization?”

Curious, he posed the same question to three other vendors he was speaking with.

One vendor’s sales consultant opened Doubao, searched “What is vectorization?”, captured the answer as a screenshot, and sent it directly to the client.

The client was furious: “Seriously? If I wanted Doubao’s answer, I’d ask Doubao myself. Why send me a screenshot?”

He circled back to us.

Our rep explained why vectorization matters in GEO: how it enables semantic search, powers dynamic content matching, affects crawl budget allocation, and influences ranking stability across regional queries. He grounded it in the client’s actual site architecture and traffic patterns.

The contrast was immediate. The client felt a clear difference in depth, respect, patience, and domain fluency.

The deal closed. Nearly one million RMB.

This case is emblematic. The technical question also served as a capability audit. The client wanted to know whether the salesperson truly understood the subject.

In the AI era, sales can no longer rely on information asymmetry. Your prospect has ChatGPT, DeepSeek, and Doubao. If your answer mirrors theirs, they’ll soon ask: Why do I need you?

The highest-value salespeople deliver incremental insight: context, trade-offs, implementation caveats, and strategic implications that require deep domain grounding. Every question becomes a quiet test. The client’s final verdict is: Does this company truly understand?

Why Your Website Matters More Than Ever

In the AI era, your official website is your organization’s authoritative information foundation in the digital world.

For AI, your website is the primary source: To understand your company, AI agents rely first on your site, the only place with first-party, attributable, auditable facts.
It serves dual audiences: Humans and machines. Tomorrow’s site must be legible to people and parseable, indexable, and citable by AI crawlers, search engines, and autonomous agents.
It shapes AI’s “answer” about you: A neglected, outdated, or fragmented site forces AI to fall back on third-party sources, news, wikis, social chatter, risking misrepresentation, underestimation, or total omission of your true capabilities.
It’s becoming a service layer: As agents take on more concrete tasks (e.g., booking demos, retrieving pricing, submitting support tickets), your site evolves into a transactional interface and an entry point for AI-mediated actions.

Yet most corporate sites remain optimized solely for human visitors and legacy SEO. They lack structured data, semantic markup, coherent content hierarchies, and machine-readable documentation. That gap is widening fast, and closing it will be the next major upgrade cycle for enterprise web strategy.

A Brief History of Credit

Before writing or coin, credit lived in memory: sharing food today meant trusting kinship, reputation, and communal bonds to secure tomorrow’s return.
~7500 BCE, Near Eastern farming communities used clay tokens with distinct shapes for grain and livestock, turning memory into countable, transferable records.
~3200 BCE, early writing emerged alongside accounting; records became tools to fight forgetting.
~2000 BCE, Mesopotamian loan tablets formalized principal, term, interest, witnesses, and repayment, shifting credit from favor to audit.
18th c. BCE, Hammurabi’s Code distinguished deliberate default from crop failure, acknowledging that credit must allow for renewal after collapse.
Spring & Autumn period, Chinese merged “person” (rén) and “speech” (yán) into “trust” (xìn), declaring “Without trust, a state cannot stand.”
Warring States period, Shang Yang’s “wood-pole promise” publicly rewarded a citizen who moved a pole and proved that credible institutions begin with visible, verifiable acts.
Medieval Islamic trade networks used hawala (remittance letters), debt transfers, and mudarabah (profit-sharing partnerships) to move value and enforce obligations across continents.
11th c. Song Dynasty, jiaozi paper money replaced heavy iron coins and drew its support from reserves, redemption discipline, and collective confidence in the issuer.
11th c. Maghribi traders circulated agent misconduct reports across networks, making the loss of future opportunities a steeper penalty than fines for broken trust.
12th-13th c. Champagne Fairs developed merchant law and rapid arbitration, letting strangers ship goods, extend credit, and wait for payment across distant markets.
1494, Pacioli codified double-entry bookkeeping, enabling enterprises to verify, scale, and persist beyond any single founder’s lifespan.
1694, the Bank of England monetized future tax revenue, extending national credibility into long-term finance and warfare.
1748, Franklin noted that timely payments compound credit; each on-time settlement lowers the cost of the next deal.
1841, Lewis Tappan founded the first commercial credit bureau in NYC. The bureau turned local gossip into portable, archival records and made credit a transregional information business.
From 1909 bond ratings to 1989’s universal FICO score, institutions compressed people into symbols, speeding decisions and concentrating judgment power.
Post-1990s, the internet made reputation real-time: one review, one transaction history, one persistent identity now directly affects price.
2008 taught a harsh lesson: when markets delegate judgment to ratings, models, and leverage, credit collapses along interdependent balance sheets.
Post-2009, blockchain encoded trust into cryptography and shifted new questions to keys, code, oracles, governance, and real-world enforcement.
Now, in the AI era, when machines speak, create, transact, and act, we face humanity’s oldest question anew: Who is it? What has it done? What is it permitted to do? And when it fails, who bears responsibility? As AI shifts from “answer-giver” to “agent acting on your behalf,” what grounds our delegation of authority?

This 10,000-year arc reveals a pattern: Memory makes promises possible. Writing makes them durable. Institutions make them enforceable. Markets give them price. Data makes them quantifiable. AI systems that speak and act require promises to be machine-verifiable and machine-accountable.

GEOHub: An Open GEO Super-Skill

We’re open-sourcing GEOHub, a unified, production-ready Skill for geo-optimization.

GEOHub integrates research, diagnostic analysis, and content generation into a single capability registry, with built-in measurement and SEO interfaces for centralized management and scalable extension.

GitHub repo, design principles, capability visualization reports, and reference links are in the comments.

Core Skill architecture:

Includes a capability registry, intent router, workflow orchestrator, modular sub-skills, unified output protocol (JSON Schema), evidence ledger, quality reporting, CLI, testing suite, and packaging system.
Features a unified entry point that understands both English and Chinese GEO queries, then selects the optimal skill(s) from the registry. Single-intent requests route to minimal viable skills; multi-stage requests (e.g., “diagnose my brand’s GEO gaps then produce content”) trigger stable, validated workflows:
- Problem discovery → Brand diagnosis
- Problem discovery → Content production
Covers 14 GEO/SEO task types: issue/opportunity discovery, comparative analysis, assessment, action planning, brand/site/page diagnostics, and seven content modes, titles, explainers, comparisons, rankings, page blueprints, content optimization, and reader-friendly rewrites, as well as full-site SEO strategy.
Validated and iterated using the latest global GEO research datasets and 54 peer-reviewed GEO papers, ensuring alignment between practical implementation and academic rigor.

GEOHub resources:

GitHub repository: GEOHub
Design principles and capability visualization report: doc.laoyao.cn
GEO dataset and paper repository: geo-citation-lab
Collection of GEO-specific skills: yao-geo-skills
Lightweight, one-line website SEO tool: qiaomu-seo
Reference architecture (Waza skill framework): Waza

China’s GEO Ecosystem: Business Model Landscape Report

We’ve compiled the China GEO Ecosystem Business Model Landscape Report.

Maps the domestic GEO ecosystem across 7 layers: Entry points, information sources, data infrastructure, core technologies, execution services, transaction platforms, and governance frameworks.
Identifies and analyzes 19 distinct GEO business models, benchmarked horizontally by maturity × commercial ceiling.
Highlights near-term cash flow leaders: GEO consulting & training, GEO operations outsourcing, and GEO content distribution.
Flags long-term strategic bets: GEO-native data platforms and GEO operating systems (GEO OS).

Full report: doc.laoyao.cn