The Future of Business as AI Capabilities Evolve

Key Narrative

We are in the early stages of a transformation in what businesses can do and how they’re structured. AI capabilities are improving rapidly—not just in language and image generation, but in reasoning, planning, and executing complex tasks. This post attempts to think through the implications for business over the next 3-5 years.

The core argument: AI will compress the relationship between company size and capability. Small teams will be able to do what once required large organizations. But this doesn’t mean all large companies disappear—it means the reasons for scale shift. Distribution, trust, capital, and data advantages remain. Pure labor leverage does not.

Outline

I. Introduction: The Capability Explosion

• Where we are today (early 2026)
• The rate of change: what’s improved in 24 months
• Why this time might be different
• Framing: not “will AI matter?” but “how, and how fast?”

II. What AI Can Do Now (and Soon)

A. Current Capabilities (2026)

• Language: writing, summarization, analysis, translation
• Code: generation, review, debugging, documentation
• Reasoning: multi-step problem solving (improving rapidly)
• Vision: image understanding, document processing
• Voice: transcription, generation, real-time interaction
• Agents: task execution with supervision

B. Near-Term Capabilities (2027-2028)

• More reliable agents: less hallucination, better tool use
• Longer context, better memory
• Specialized models for domains: legal, medical, engineering
• Multimodal native: text + image + audio + video + code
• Improved planning and decomposition

C. What Remains Hard

• Novel research at the frontier
• Physical manipulation (robotics lag)
• Deep context requiring years of experience
• Judgment in ambiguous, high-stakes situations
• Creativity that’s truly new (vs. recombination)

III. Implications for Business Structure

A. The Shrinking Team

• The 10-person company doing 100-person work
• Case study: software development with AI
• Case study: content creation at scale
• Case study: customer service without people
• Where the leverage is highest

B. The Remaining Reasons for Scale

1. Distribution: Reaching customers still requires presence
2. Trust: Brand, reputation, relationships
3. Capital: Physical assets, R&D investment
4. Data: Proprietary training data, feedback loops
5. Regulatory capture: Compliance as moat

• What large companies still do better

C. The Hollowing Out

• Middle layers of organizations most affected
• The manager’s job: coordination, or something else?
• Knowledge work that becomes automated
• The bimodal workforce: highly skilled or highly automated

IV. Industry-Specific Trajectories

A. Software & Technology

• Development velocity increases dramatically
• Testing, documentation, maintenance automated
• Small teams compete with large ones
• The open-source acceleration

B. Professional Services

• Legal: document review, contract drafting, research
• Consulting: analysis, deck creation, knowledge synthesis
• Accounting: audit, tax, compliance
• Medicine: diagnostics, documentation, research
• The partnership model under pressure

C. Media & Content

• Writing, video, audio production costs collapse
• Personalization at scale
• The authenticity premium
• Human curation becomes valuable

D. Manufacturing & Industrial

• Design and engineering acceleration
• Predictive maintenance (my company’s focus)
• Quality control and optimization
• Robotics: the lagging frontier

E. Finance

• Trading already automated
• Analysis and research
• Customer-facing automation
• Risk and compliance

V. Strategic Implications

A. For Startups

• Lower capital requirements for some businesses
• Faster iteration, faster learning
• But: defensibility harder (others can replicate)
• Focus on data, distribution, network effects

B. For Large Companies

• Talent leverage changes
• The build-vs-buy calculus shifts
• Organizational redesign needed
• Risk: incumbency advantages erode

C. For Workers

• Skills that remain valuable
• Skills that become less valuable
• The adaptation challenge
• New roles that emerge

D. For Investors

• What drives returns in an AI-augmented world?
• Capital intensity changes
• Margin expansion vs. competition effects
• The timing problem

VI. What I’m Less Sure About

A. Pace of Change

• Technical progress vs. adoption speed
• Regulatory intervention
• Social/cultural resistance
• Integration challenges

B. Employment Effects

• Net job creation vs. destruction
• Timing: how fast?
• Geographic distribution
• Social stability implications

C. Concentration Effects

• Does AI advantage accrue to large or small?
• Winner-take-all dynamics?
• New monopolies or more competition?

D. Specific Predictions

• Which companies win/lose
• Timing of capability milestones
• Regulatory outcomes

VII. What My Company Is Doing

A. AI in Industrial Settings

• Predictive maintenance at scale
• Reliability engineering automation
• Where the value is clear today

B. How We Use AI Internally

• Development velocity
• Customer support
• Analysis and reporting
• What’s working, what isn’t

C. What We’re Watching

• Model capabilities we need
• Build vs. API decisions
• The talent market

VIII. Framework for Thinking About This

A. Questions to Ask

1. What tasks in your business can be decomposed into steps?
2. Which of those steps can AI do today? In 2 years?
3. What remains that requires human judgment, trust, presence?
4. How do your competitive advantages change?
5. What new capabilities become possible?

B. Mental Models

• The technology adoption curve (but faster)
• Complements vs. substitutes
• Platform shifts (mobile analogy, but bigger)
• The innovator’s dilemma, applied

IX. Conclusion

• The change is real and significant
• The timeline is uncertain but probably faster than expected
• Adaptation is required, not optional
• The companies that figure this out early win

Suggested Sources

Current State

• Anthropic, OpenAI, Google research blogs
• AI benchmark tracking (MMLU, etc.)
• The Information, Bloomberg on AI business
• Stratechery (Ben Thompson) for strategy analysis

Economic Analysis

• Erik Brynjolfsson on AI economics
• Daron Acemoglu on labor displacement
• Goldman Sachs, McKinsey AI economic impact reports
• NBER working papers on AI and productivity

Business Strategy

• A16Z AI writing
• Elad Gil’s writing on AI startups
• Sequoia “AI 50” analysis
• Y Combinator batch trends

Skeptical Views

• Gary Marcus on AI limitations
• Melanie Mitchell on AI hype
• Emily Bender on language models

Historical Analogs

• Brian Arthur on technology adoption
• Carlota Perez on technological revolutions
• Smil on energy transitions (for pacing)