Main Site

Brand OS v1.5 · Revised Whitepaper

Brand OS v1.5: A Brand Operating System for the AI Agent Era.

AI Agent-Oriented Brand Operating System

From helping AI work within the brand system to governing the broader AI toolchain.

Read the introduction

As AI generation tools, agent platforms and automated workflows enter day-to-day enterprise operations, the way brands are produced is changing. Organizations need an upstream system that can coordinate tools, constrain outputs, review results and turn approved work into governed brand assets.

Brand OS v1.5 places Human Decision Authority at the top, treats the Brand Kernel as the governing source, and uses semantic systems, visual tokens, Agent Protocols and the Tool Invocation Layer to translate human judgment into controlled execution.

It provides a framework for moving from fragmented AI usage toward a more governed operating model — with explicit judgment, clear boundaries, defined protocols, review mechanisms and reusable assets across multiple agents, tools, departments and touchpoints.

Key Concepts

Explore the Seven-Layer Architecture View the 90-Day Roadmap

01 · EXECUTIVE SUMMARY

Three judgments: AI tools handle production; Brand OS governs it.

v1.4 addressed how AI works within the brand system. v1.5 goes further and addresses how one brand system governs the whole AI toolchain.

01 · Fragmenting

The AI Toolchain Is Fragmenting

Content, visual design, video, agent, automation and coding tools are developing into distinct ecosystems. Organizations are increasingly likely to use multiple AI Agents and tools in parallel.

02 · Drift

Brand Drift Is Accelerating

AI raises production efficiency, and it also amplifies semantic drift, visual drift, claim risk and asset contamination. The faster the output, the faster the drift; the greater the capacity, the harder the governance.

03 · Upstream

Brand OS Becomes the Upstream Layer

It defines direction, boundaries, permissions, invocation relationships, review mechanisms and learning loops, bringing dispersed AI tools together under one set of brand rules.

7Architecture Layers
10AI Agent Types
3Invocation Stages
5Review Dimensions
90Day Roadmap

02 · WHY BRAND OS v1.5

More AI tools, less brand consistency.

Tools execute specific tasks. A system defines direction, boundaries, permissions, invocation relationships and review rules.

CURRENT · MORE TOOLS, LESS CONSISTENCY

When Multiple AI Tools Work Independently

  • Copy Agentseach write in their own voiceFragmented messaging
  • Design Agentsdrift in styleVisual drift
  • Sales Agentsmay overpromiseCredibility risk
  • Video Agentsfavor spectacle over narrativeNarrative loss
  • UI Agentslook polished but convert weaklyCommercially disconnected
  • Automation Agentspush flawed content to more touchpointsErrors at scale

BRAND OS · THE GOVERNING SYSTEM

The Governance Mechanisms Brand OS Establishes

  • Establish Human Decision Authority
  • Establish the Brand Kernel
  • Establish the Visual Token System
  • Establish Agent Protocols
  • Establish the Tool Invocation Layer
  • Establish Consistency Scoring
  • Establish the Governance & Learning Loop

03 · BRAND OS EVOLUTION

Every version responds to a change in the brand environment.

Brand OS is a brand operating system that evolves continuously.

    • Parameterized judgment
    • Brand Constitution
    • Aesthetic Constitution
    • Core Narrative
    • Worldview Framework
    • Website knowledge base
    • GEO content structure
    • Authoritative first-party source
    • Citation-ready semantics
    • Case proof pages
    • Brand Context Compressor
    • Brand Semantic System
    • AI Agent Role Matrix
    • AI Aesthetic Constitution
    • Drift monitoring
    • Human Decision Authority
    • Visual Token System
    • Tool Invocation Layer
    • Agent-to-Tool Mapping
    • Brand Mission Control
    • Consistency Scoring

What v1.5 addresses: how Brand OS orchestrates and constrains the work of an AI toolchain — using Brand OS as the governing system to coordinate different Agents, invoke different tools and constrain different outputs, while strategic judgment remains with the organization.

04 · DEFINITION & AUTHORITY

A foundational system that helps the brand retain sovereignty across an AI toolchain.

Full definition

Brand OS v1.5 is a brand operating system designed for an AI Agent environment. It is grounded in the Brand Kernel, bounded by the Brand Constitution, supported by semantic and visual systems, organized through Agent Protocols, executed through the Tool Invocation Layer, and governed through review mechanisms and learning loops. It helps organizations build a governing brand system that can be used by people, referenced by AI, executed through tools, reviewed by systems and maintained by the organization over time.

  1. 01 · Highest Authority

    Human · Strategic Judgment

    Human Decision Authority

    Owners, executives, brand leaders and advisors hold final strategic judgment. This is the highest level of authority in Brand OS, and introducing AI does not transfer it.

  2. 02 · Orchestration

    Brand OS · Orchestration Authority

    System Orchestration

    Brand OS defines tasks, boundaries, permissions, inputs, outputs and review standards, and coordinates Agents and tools. It does not make strategic trade-offs; it makes human judgment executable in a consistent way.

  3. 03 · Bounded Execution

    AI Agent · Bounded Execution

    Controlled Execution

    Agents analyze, generate, simulate and execute within the constraints of the brand system. Their output enters the brand asset library only after review.

05 · SEVEN-LAYER ARCHITECTURE

Seven layers: from human judgment to governed AI execution.

The full path from brand assets to tool invocation, and from generated output to governed review. Select any layer to read its description.

  1. Who holds the highest authority over the brand. AI may assist, execute, simulate, generate and review; final strategic judgment remains with the organization. This is the highest level of authority in Brand OS.

  2. The brand foundation every AI Agent works from: positioning, value proposition, core narrative, worldview framework, Brand Constitution, Aesthetic Constitution and differentiating evidence. It is the governing source context for AI Agents.

  3. How different AI Agents receive usable brand context at different depths: 30-second, 3-minute, 15-minute and full-context versions of the Brand Context Compressor, plus FAQ, sales messaging and multilingual semantic equivalence.

  4. Which brand assets AI tools can draw on: logo, brand colors, typography, graphic elements, imagery, layout, Visual Exclusion Rules, content assets, evidence assets and Prompt Assets. The focus here is the Visual Token System.

  5. What each AI Agent does and does not do. Every Agent has a defined role, permissions, input assets, output standards and review mechanisms.

  6. How Brand OS works with external AI tools through one consistent interface — visual and video generation, design collaboration, agent platforms, workflow automation, AI coding and more. Tools keep evolving and can be swapped at any time. The governing principle: Agents remain stable; tools can be replaced.

  7. How AI-generated output is reviewed, archived, corrected and improved: Consistency Scoring, human review, asset archiving, tool replacement and the learning loop. Generated output does not automatically become an approved brand asset.

06 · VISUAL TOKEN SYSTEM

Turn identity guidelines into a visual language AI can execute.

Every visual generation and design tool — including built-in brand kits and generation based on visual identity guidelines — works from one structured set of visual tokens, reducing the style drift of ad hoc prompting. Hover over any token line to see what it controls.

brand.tokens.json

// Brand Kernel · color tokens
"color.brand.blue"     : "#0A56F9",
"color.brand.navy"     : "#07152D",
"color.ratio.primary"  : "60%",
"color.forbidden"      : ["neon","gradient-stack"],
// typography
"type.display"         : "Noto Sans SC / 600",
"type.body"            : "PingFang SC / 400",
"type.scale"           : [44, 28, 18, 14],
// layout · image
"layout.grid"          : 12,
"layout.whitespace"    : "generous",
"image.light"          : "soft-blue",
"image.material"       : "restrained-metal",
"image.forbidden"      : ["cyberpunk","over-glow"]

Traditional description → AI-executable expression

Visual ElementTraditional DescriptionAI-Executable Expression
Brand ColorThe primary color is deep blueHEX values, usage ratios, prohibited combinations
TypographyUse a sans-serif faceType hierarchy for headings, body, Latin text and numerals
Graphic SystemSupporting graphicsPermitted shapes, scale, opacity and combination rules
ImageryA sense of technologyLighting, materials, composition, camera and prohibited styles
LayoutClean and refinedInformation density, whitespace ratio, grid and hierarchy rules

07 · AGENT PROTOCOL · 10 AGENTS

Ten AI Agents, each with a role and a boundary.

Every Agent has a defined role, permissions, callable assets, core risk and permission principle — moving the organization from “we use AI for branding” to “we use Brand OS to coordinate AI Agents with clearly defined responsibilities.”

Showing all 10 agent types

  • Strategy AgentAnalysis & Positioning

    strategy

    Core task: Support industry, competitor, trend and positioning analysis

    Callable assets: Brand Kernelindustry researchcompetitor research

    Core risk: Misreading strategic direction

    Permission principle: Advisory only; does not make final decisions

  • Kernel AgentBrand Kernel Definition

    strategy

    Core task: Organize brand positioning, propositions, Brand Constitution, Aesthetic Constitution and narrative

    Callable assets: Brand interviewsstrategy documentsmanagement judgment

    Core risk: Turning opinions into apparent conclusions

    Permission principle: Requires human review

  • Semantic AgentLanguage & Messaging

    semantic

    Core task: Generate website copy, sales messaging, FAQ responses and article content

    Callable assets: Brand Semantic SystemEvidence LedgerFAQ

    Core risk: Semantic drift and generic language

    Permission principle: Claims must reference supporting evidence

  • Design AgentVisual Generation

    visual

    Core task: Generate key visuals, posters, images and brand applications

    Callable assets: Visual identityVisual TokensAesthetic Constitution

    Core risk: Visual drift and style contamination

    Permission principle: Review before use

  • Video AgentMotion & Film

    visual

    Core task: Generate dynamic visuals, brand films and video covers

    Callable assets: Visual Tokensscriptsimagery style

    Core risk: Spectacle over narrative

    Permission principle: Important videos require human review

  • UI AgentInterfaces & Prototypes

    visual

    Core task: Generate websites, product interfaces and admin prototypes

    Callable assets: Brand visualscontent structureinteraction principles

    Core risk: Visually polished but commercially disconnected

    Permission principle: Must align with the intended business journey

  • Sales AgentCustomer Conversations

    semantic

    Core task: Answer customer questions and support early-stage communication

    Callable assets: FAQservice boundariescase evidence

    Core risk: Overpromising

    Permission principle: Must not promise outcomes beyond approved claims

  • Knowledge AgentKnowledge Sources

    semantic

    Core task: Organize website, cases, documents and the knowledge base

    Callable assets: WebsitearticlescasesFAQ

    Core risk: Using outdated materials

    Permission principle: Must follow version-control rules

  • Workflow AgentProcess & Approvals

    workflow

    Core task: Connect tasks, approvals, notifications and archiving

    Callable assets: Process rulespermission matrix

    Core risk: Automation errors

    Permission principle: Critical actions require human approval

  • Audit AgentConsistency Review

    governance

    Core task: Review semantic, visual, evidence and context consistency

    Callable assets: Brand ConstitutionAesthetic Constitutionreview checklists

    Core risk: Unclear review standards

    Permission principle: Provides recommendations; does not replace final human judgment

08 · TOOL INVOCATION LAYER

Input packages first, then semi-automation, then API integration.

Brand OS supplies brand context, rules, task inputs and review standards upstream; the generation itself is still performed by the corresponding tools.

Brand OS produces an AI Design Agent Input Package, which the user loads into whichever visual generation or design tool the organization currently uses.

  1. Generate the Brand Input Package
  2. Load it into the visual tool
  3. Generate within token constraints

Task briefs, prompts and asset packages are generated in the page or system, then transferred by the user into the relevant tool for execution. All inputs originate from Brand OS.

  1. Generate the task brief
  2. Generate prompt + asset package
  3. Transfer to the execution tool

The user selects a touchpoint task; Brand OS loads the Brand Kernel, semantic system and visual tokens; the Design Agent creates a brief; the Tool Invocation Layer calls the external tool; the Audit Agent scores the output; and after human review the approved asset enters the library. This describes the target architecture rather than a currently deployed integration — on this basis a Brand Mission Control prototype can centralize tasks, assets and reviews in one interface.

  1. Select the touchpoint task
  2. Load Kernel / Tokens → create the brief
  3. Invoke the tool → Audit scoring
  4. Human review → asset enters the library
  • 01Agents remain stable; tools can change
  • 02Protocols remain stable; interfaces can evolve
  • 03The Brand Kernel remains continuous; execution tools can evolve

09 · AGENT-TO-TOOL MAPPING

Agents stay stable; tools can be replaced.

The eight Agent types for which tool directions are currently defined. Select an Agent to see example tools; press “Swap tool” to see the Agent and the role of Brand OS stay the same. Tool names are examples from today's ecosystem and will change as tools evolve — Brand OS defines Agent responsibilities and brand standards, not specific products. Actual configuration depends on the organization's tool environment and the scope of the engagement.

Example tools · Swappable

    10 · CONSISTENCY SCORING & GOVERNANCE

    AI can generate, but nothing enters the asset library unreviewed.

    Every significant AI output is reviewed across five dimensions. Drag the sliders below to see how the framework arrives at a usage recommendation.

    92

    Review focus: whether the target customer, competitive position and value proposition have been diluted.

    86

    Review focus: whether terminology, tone and sentence structure align with the existing semantic system.

    78

    Review focus: whether color ratios, type hierarchy, graphic and imagery styles stay within bounds.

    70

    Review focus: whether claims are bound to verifiable facts in the Evidence Ledger.

    88

    Review focus: whether touchpoint, stage and audience match, and whether the expression is over- or under-stated.

    83/ 100 demo score

    Revise before use

    1. 90–100Eligible for the brand asset library
    2. 75–89Revise before use
    3. 60–74Exploratory reference only
    4. < 60Not recommended for use

    Scores should only be generated against specific outputs, explicit review criteria and actual project evidence. This whitepaper presents the scoring framework only and shows no real company scores; this is an interactive demonstration.

    1. 01Consistency ScoringFive Review Dimensions

      Every significant AI output is reviewed across five dimensions.

    2. 02Human ReviewHuman-in-the-Loop

      Positioning, key visuals and outward-facing English copy require human approval.

    3. 03Asset ArchivingBrand Asset Library

      Assets are managed in four tiers: core, contextual, exploratory and deprecated.

    4. 04Tool Replacement ProtocolStable Agents, Swappable Tools

      Tools can change without destabilizing Brand OS, provided the affected scope is defined.

    5. 05Learning LoopFeedback into the Kernel

      Feedback flows back to update the kernel, semantics, tokens, protocols and invocation rules.

    11 · REAL PAIN POINTS

    Seven recurring pain points, seven matching mechanisms.

    Brand OS v1.5 is designed to help organizations maintain clarity and consistency across increasingly complex AI tool environments.

    1. Many AI tools, but increasing brand inconsistency

      No upstream orchestration system

      Brand Mission Control + Agent Protocols

    2. AI generates visuals fast, but style may drift

      No Brand Kernel or Visual Tokens to work from

      AI Design Agent Input Package + Aesthetic Constitution

    3. Sales and service AI give inconsistent answers

      No Brand Semantic System

      FAQ + sales messaging + semantic context

    4. AI copy reads well but lacks evidence

      No Brand Evidence Ledger

      Claim Control + Evidence Binding

    5. Concern that AI may replace strategic judgment

      Unclear authority boundaries

      Human Decision Authority

    6. Different tools produce inconsistent output formats

      No output standards

      Tool invocation standards + export specifications

    7. No reliable way to evaluate AI-generated brand content

      No review standards

      Brand Consistency Scoring system

    12 · 90-DAY ROADMAP

    90 days, from diagnosis to operation.

    Six phases establish a first working interface for Brand OS and a review mechanism. The realistic sequence: input packages first, then semi-automated workflows, then deeper API integration. Select a phase to see its goal, key actions and deliverables.

    1. Day 0–15 · Goal: identify the breakpoints in the organization's current brand system and AI usage.

      Key Actions

      • Review brand positioning and the value expressed on the website
      • Inventory the AI tools currently in use
      • Identify AI outputs already creating brand inconsistency
      • Review case evidence and FAQ coverage

      Deliverables

      • Brand System Diagnostic Report
      • AI Toolchain Current-State Assessment
      • Initial Optimization Priorities

    2. Day 15–30 · Goal: establish the highest decision layer of Brand OS.

      Key Actions

      • Clarify brand positioning and value proposition
      • Rebuild the core narrative
      • Define the Brand Constitution and Aesthetic Constitution
      • Define which decisions require human review and approval

      Deliverables

      • Brand Positioning + Value Proposition System
      • Brand Constitution + Aesthetic Constitution
      • Human Decision Authority Matrix

    3. Day 30–45 · Goal: enable the brand to be interpreted and used more consistently by AI-assisted workflows.

      Key Actions

      • Build the Brand Context Compressor
      • Build the Brand Semantic System and FAQ
      • Analyze visual identity documentation and extract Visual Tokens
      • Define Visual Exclusion Rules

      Deliverables

      • Brand Context Compressor + Brand Semantic System
      • Visual Token System
      • Visual Exclusion Rules

    4. Day 45–60 · Goal: define each AI Agent's role, permissions and tool selection.

      Key Actions

      • Define the Agent Role Matrix
      • Build the Agent-to-Tool Mapping
      • Define callable assets and prohibited behaviors
      • Build Prompt Base Templates

      Deliverables

      • AI Agent Role Matrix
      • Agent-to-Tool Mapping
      • Agent Permission Matrix + Tool Replacement Protocol

    5. Day 60–75 · Goal: prove out the visual AI workflow first — suggested touchpoints are the website banner, social media covers and business development posters.

      Key Actions

      • Generate the AI Design Agent Input Package
      • Load it into the AI generation tools the organization uses
      • Review the output with Consistency Scoring
      • Archive the approved assets

      Deliverables

      • AI Design Agent Input Package
      • Touchpoint Generation Brief
      • Visual Output Review Checklist

    6. Day 75–90 · Goal: establish a first working interface for Brand OS together with a retrospective mechanism.

      Key Actions

      • Design the Brand OS UI prototype
      • Build the asset library, task center and review center
      • Establish version records
      • Establish monthly review and quarterly retrospective mechanisms

      Deliverables

      • Brand Mission Control UI Prototype
      • Brand Asset Library / Task Center / Review Center structure
      • Monthly Review + Quarterly Retrospective Mechanism

    The 90 days illustrate a reference implementation cadence for Brand OS v1.5, not a contractual delivery guarantee. Actual timing depends on existing brand assets, organizational complexity, AI tool environment, data readiness and depth of systems integration.

    13 · DELIVERABLES

    26 deliverables, one operational set of brand assets.

    Designed for human use, AI reference, tool-based execution, organizational review and ongoing maintenance.

    A

    Brand Kernel

    • Brand Kernel Definition
    • Brand Constitution
    • Aesthetic Constitution
    • Human Decision Authority Matrix
    • Value Proposition System
    • Core Narrative Framework

    B

    Semantics & Evidence

    • Brand Context Compressor
    • Brand Semantic System
    • FAQ Standard Responses
    • Sales Messaging System
    • Brand Evidence Ledger
    • English Language Guidelines

    C

    Agents & Tools

    • AI Agent Role Matrix
    • Agent-to-Tool Mapping
    • AI Design Agent Input Package
    • AI Prompt Base Templates
    • AI Claim Control Matrix
    • AI Brand Review Checklist
    • Brand Mission Control UI Prototype

    D

    Governance & Operations

    • Brand Consistency Scorecard
    • Brand Permission Matrix
    • Tool Replacement Protocol
    • Semantic / Visual Drift Monitoring
    • Brand Decision Log Template
    • Monthly Review Mechanism
    • Quarterly Retrospective & Update Mechanism

    14 · WHO IS BRAND OS v1.5 FOR

    Built for organizations that intend to keep brand sovereignty in the AI era.

    Brand OS v1.5 suits organizations that treat the brand as long-term operating infrastructure.

    STRONG FIT · BEST ALIGNED WHEN

    • Organizations already using AI tools but experiencing inconsistent brand outputs
    • Growth companies moving from visual refresh toward system-level brand development
    • Manufacturing, technology, clean energy and B2B organizations with complex capabilities and long decision chains
    • Companies preparing for international expansion that need to reorganize brand semantics and evidence
    • Organizations across multiple departments, business lines and touchpoints that need permission and version governance
    • Companies planning an AI brand advisor or website AI assistant that need a standardized knowledge source first

    NOT YET · MAY BE PREMATURE

    • Organizations that only need a logo
    • Organizations that only want a short-term visual refresh
    • Organizations that do not want to participate in strategic brand decisions
    • Organizations whose product, service and business foundation remains highly uncertain, for which a complete Brand OS is not the immediate priority
    • Organizations that only want a one-off document rather than a long-term system

    Brand OS cannot replace product capability, supply chains, business models or organizational capability. Its role is to align brand judgment, expression and execution on top of those fundamentals.

    15 · XINMING VIEW

    Make the brand the governing context, rule set and constraint for every AI Agent.

    Six judgments on brand governance in the AI era.

    1. 01The stronger AI tools become, the more a brand needs a stable governing system.
    2. 02The faster visual production becomes, the clearer the rules need to be: AI tools speed up visual production; Brand OS keeps the brand clear.
    3. 03AI can assist judgment; strategic responsibility remains human.
    4. 04Tools can change; the Brand Kernel needs continuity.
    5. 05Agents and interfaces can evolve; governance rules need stability.
    6. 06Part of future brand competition will depend on how clearly and consistently AI can interpret and execute the brand.
    Visit the Xinming Main Site

    16 · FAQ

    Frequently Asked Questions

    Eight Questions on Brand OS v1.5

    Eight recurring questions about Brand OS v1.5.

    01How Is Brand OS Different from a Traditional Visual Identity Manual?

    A traditional visual identity manual focuses primarily on whether the brand looks consistent, and it is written for people to read and apply. Brand OS focuses on long-term consistency across the organization, AI Agents and the toolchain. It organizes the Brand Kernel, semantics, visual tokens, Agent Protocols, invocation interfaces and governance mechanisms into a governing brand system that people can execute, AI Agents can reference and systems can review.

    02Why Does Brand OS v1.5 Emphasize AI Agents?

    Organizations are increasingly likely to use multiple AI Agents and tools in parallel. v1.5 defines each Agent's role, permissions, input assets, output standards and review mechanisms, so that different AI systems work within one brand system instead of each writing and designing on its own terms.

    03If a Company Already Uses Many AI Tools, Does It Still Need Brand OS?

    Tools execute specific tasks; a system defines direction, boundaries, permissions, invocation relationships and review rules. The more AI tools an organization uses, the more it needs an operating layer above them to coordinate their work. Brand OS unifies context, permissions, rules and review so that dispersed tools operate within one brand framework.

    04What Problem Does the Visual Token System Solve?

    It translates conventional visual identity documentation into a structured visual language that AI tools can reference: brand colors are expressed as HEX values and usage ratios, typography as a hierarchy, graphics as form and combination rules, imagery as lighting and camera treatment, and layout as grid and hierarchy. Whatever visual generation or design tools an organization uses can then work from the same defined visual tokens, helping reduce the style drift that comes from ad hoc visual generation.

    05What Role Do AI Generation Tools Play in Brand OS?

    They act as the execution environments for the relevant Agents — covering image, video, layout, interface, copy and automation capabilities — and handle visual generation, touchpoint extension and campaign execution. These tools evolve quickly, and more of them now support imported brand kits and generation based on visual identity guidelines. Brand OS is therefore not tied to any specific tool: it sits upstream as the brand's single source of truth, providing the Brand Kernel, Aesthetic Constitution, visual tokens, task briefs, permission boundaries and review standards, and translating them into whatever input formats each tool currently supports, such as brand kits, style references or prompt templates. Tools can be upgraded or replaced; the brand standards and the review gate stay the same.

    06Will Brand OS Replace Designers or Brand Leaders?

    No. Brand OS explicitly preserves Human Decision Authority: key judgments such as brand positioning, value propositions, market trade-offs and final visual direction remain human responsibilities. AI assists with analysis, generation and review, allowing designers and brand leaders to focus more of their attention on judgment and governance.

    07What Types of Companies Are Best Suited to Brand OS?

    It suits organizations already using AI tools whose brand outputs have started to diverge, growth companies undergoing brand transformation, manufacturing, technology, clean energy and B2B organizations with complex capabilities and long decision chains, and companies preparing for international expansion or operating across many touchpoints. The key condition is a willingness to treat the brand as long-term operating infrastructure.

    08What Can Be Accomplished in 90 Days?

    The 90 days are organized into six phases: brand system and AI toolchain diagnosis; Brand Kernel and decision authority; semantic system and visual token development; Agent Protocols and tool mapping; an AI Design Agent input package pilot; and finally a Brand Mission Control prototype with a learning loop. Input packages come first, then semi-automated workflows, and only then deeper API integration. The 90-day structure is a reference implementation cadence proposed by this whitepaper; actual timing depends on the organization's brand assets, organizational complexity, tool environment and depth of systems integration.