01 / Architecture Shift
The Value Stack Inverts
In the SaaS era, value flowed upward to application vendors. AI collapses the application layer into infrastructure. The intelligence layer moves down; the value capture point shifts from user interface to agent orchestration.
SaaS-Era Stack
User Interface
Value Capture
Application Layer
Differentiated
Infrastructure
Commodity
Shift
AI-Native Stack
Interface (Exception Handling Only)
Minimal
Agent Orchestration
Value Capture
Infrastructure + Intelligence
Differentiated
Key insight: The intelligence layer collapses into infrastructure. Value that previously flowed to application vendors now accrues to whoever controls the AI-optimized infrastructure stack.
02 / SaaS Incumbents
The Innovator's Dilemma Crystallized
SaaS companies face brutal asymmetry: their strengths are substantial but their weaknesses are architectural. They cannot be patched with new features or incrementally evolved. They require fundamental rebuilds.
Strengths (Substantial)
- Installed base of millions of paying, integrated users
- Deep workflow understanding of how work gets done
- Years of operational data showing process execution
- Executive-level customer relationships and renewal inertia
Weaknesses (Structural)
- Architecture optimized for human operation, not agents
- Revenue model dependent on seat expansion
- Org structure built for shipping interface features
- Technical debt optimized for the wrong paradigm
Why "Add AI Features" Fails
Three Strategic Paths for Incumbents
Path 01
Bet Everything
Rebuild architecture from scratch. Accept near-term revenue cannibalization. Retrain or replace organizational capabilities.
High-Risk / High-Reward
Path 02
Hybrid Strategy
Maintain core SaaS while building AI-native alongside. Salesforce (Agentforce) and Workday (AI agents) are attempting this path.
Most Common
Path 03
Defend & Harvest
Focus on regulated industries and complex workflows where AI substitution is slower. Maximize cash flow from existing base.
Managed Decline
03 / Infrastructure Giants
Structural Advantages That Compound
Microsoft, Google, and Amazon control the full stack. They embed intelligence at every layer and capture value at points others cannot access. Their advantages compound through cost, data, and switching costs.
Why These Advantages Compound
The Application Vendor Squeeze: HCM apps get squeezed by Microsoft's HR Copilot. Marketing automation gets squeezed by integrated AI in productivity suites. Customer support gets squeezed by AI agents on infrastructure layers. Staying in the middle is untenable.
04 / AI-Native Vertical Players
Bypassing SaaS Entirely
The most disruptive dynamic: AI-native companies don't add AI to SaaS. They build autonomous agents that treat SaaS data as an input, not the value layer. They deliver outcomes, not features.
Harvey AI — Legal
$3B+ Valuation
Not legal practice management SaaS. An AI legal researcher that performs legal work autonomously: analyzing case law, drafting documents, conducting due diligence, and delivering strategic analysis. Uses legal databases (Westlaw, LexisNexis) as data sources; value is in autonomously analyzing them to deliver legal outcomes.
Traditional Legal Tech
Pricing Model$50-200/seat/mo
Value UnitInterface Access
FunctionCase Management
Harvey (AI-Native)
Pricing ModelOutcome-Based
Value UnitLegal Work Delivered
FunctionAutonomous Analysis
Agentic — Marketing
End-to-End Autonomous
Not Marketo, HubSpot, or Pardot with AI features. Autonomous AI agents that execute marketing workflows end-to-end: analyze performance data, generate creative assets, optimize campaigns, manage budgets, and report outcomes without human operators clicking through interfaces.
The value migration: In traditional marketing automation, value comes from helping marketers execute workflows more efficiently. In agentic marketing, the AI replaces the need for the marketing automation platform entirely.
Writer — Content Operations
AI-Native Content
AI agents that produce and manage content across systems. They don't just help writers write faster. They autonomously generate content, maintain brand consistency, adapt messaging for different channels, and optimize based on performance. Built on APIs and direct data access, not human-operated interfaces.
Architectural insight: Writer builds infrastructure for AI agents to autonomously produce content. It accesses content repositories, brand guidelines, performance data, and distribution channels through APIs optimized for autonomous execution.
Klarna — Customer Service
2.3M Conversations/Month
Klarna's AI agent handles 2.3 million customer service conversations monthly, equivalent to 700 full-time workers. Value isn't in "AI agent seats" but in conversations handled, issues resolved, and customer satisfaction maintained. The economic unit is the outcome, not the seat.
Traditional Support SaaS
PricingPer Agent Seat
ValueTicket Interface
Human FTEs700 Required
Klarna AI Agent
PricingPer Resolution
ValueAutonomous Resolution
AI Equivalent700 FTEs
Why AI-Native Verticals Win
No legacy architecture constraining AI capabilities
Business model aligned with autonomous execution from day one
Organization built around agent orchestration, not interfaces
Customer expectations set for outcomes, not features
Domain expertise that can't be replicated by generic models
Specialized agents optimized for specific industry use cases
05 / Organizational Transformation
The Coordination Layer Collapses
AI doesn't just change products; it collapses the coordination layers that traditional organizations rely on. Middle management, project managers, operations teams, and administrative staff handled coordination work that now gets embedded in infrastructure.
Traditional SaaS + Hierarchy
Executive Strategy
Middle Management (Translates Strategy)
Project Managers (Orchestrates Work)
Operations Teams (Manage Workflows)
Administrative Staff (Communication)
Information flows up, decisions flow down. Hierarchies exist primarily to coordinate work.
AI-Native + Flat Structure
Strategic Judgment & Goal-Setting
Embedded AI (Coordination & Orchestration)
Autonomous Team (2-4 people)
Autonomous Team (2-4 people)
Autonomous Team (2-4 people)
Autonomous Team (2-4 people)
Outcome-driven. Teams organize around results. AI eliminates coordination overhead.
Case Studies in Radical Flattening
IT Department Transformation
SaaS Era IT
RoleTool Purchaser
FocusVendor Management
Metric% Users Active
SkillsApp Management
AI Era IT
RoleIntelligence Architect
FocusAgent Orchestration
MetricValue Delivered
SkillsMLOps / Infra Eng.
06 / Business Model Transformation
From Seat-Based to Outcome-Based Economics
SaaS achieved the holy grail: near-zero marginal cost per customer and 80%+ gross margins. AI returns software to infrastructure economics with meaningful variable costs per interaction, compressing margins and demanding multi-dimensional monetization.
SaaS Economics
Marginal Cost~Zero
Gross Margins80%+
Revenue UnitSeat / User
Price Example$50-200/seat/mo
Scale EffectAlways Improves
AI-Native Economics
Marginal Cost$0.50-2.00/op
Gross Margins50-60%
Revenue UnitOutcome
Dimensions4-Layer Hybrid
Scale EffectCan Worsen
The Four Monetization Dimensions of AI-Native
Example: Enterprise AI Platform Pricing
Base subscription: $10K/month for platform access
Usage charges: $0.02/API call, $5/GPU hour
Capability tiers: +$5K/mo advanced models, +$15K/mo custom agents
Outcome premiums: 10% of cost savings, 5% of AI-attributed revenue
Usage charges: $0.02/API call, $5/GPU hour
Capability tiers: +$5K/mo advanced models, +$15K/mo custom agents
Outcome premiums: 10% of cost savings, 5% of AI-attributed revenue
Microsoft's 3-layer monetization: Office 365 base ($20-57/user/mo) + Copilot add-on ($30/user/mo) + Azure consumption (variable compute). Monetizes at three levels while competitors struggle at one.
07 / Strategic Verdict
Who Wins and Who Gets Left Behind
The competitive landscape rewards infrastructure players with vertical integration, AI-native verticals with deep specialization, and radically flat organizations. The timeline is measured in years, not months, but structural advantages compound inexorably.
Winners
Microsoft
Best positioned: Azure + Office 365 (300M+ users) + OpenAI partnership + enterprise relationships. Full-stack integration play.
Google
Only fully integrated AI-optimized stack from custom silicon (TPUs) to application layer (Workspace). Structural cost advantage compounds with scale.
Amazon (AWS)
Platform for enterprise AI: Bedrock model marketplace, Trainium/Inferentia custom silicon. Picks-and-shovels at unprecedented scale.
Harvey, Agentic, Writer
Deep domain specialization + outcome-based economics + proprietary data moats in specific verticals.
Anthropic, Perplexity
Radically flat orgs (2-4 person teams). Permanent beta organizational structures that eliminate coordination overhead.
Structural Losers
SaaS Incumbents Defending Legacy
Death by a thousand cuts as AI-native competitors capture new customers and infra players bundle replacements.
App Vendors Without Infra Control
Squeezed from above (infra players) and below (vertical AI-natives). Middle layer becomes untenable.
Traditional Hierarchical Orgs
Pay costs of both coordination methods (hierarchy + AI infra) while capturing benefits of neither.
Seat-Based Pricing Companies
AI agents don't occupy seats. Success in AI directly reduces addressable market for core SaaS product.
The bottom line: This isn't a story of "better products winning." It's a story of architectural advantages compounding, business model misalignment destroying profitable companies, and organizational structures either adapting or becoming obsolete. The question isn't whether this transformation happens. It's which companies successfully navigate it.