How AI is Changing Enterprise
The conversation you shared is a wide-ranging and insightful reflection on the state of AI in the enterprise, featuring industry veterans like Aaron Levie (CEO of Box) and partners from Y Combinator. Here's a clear, structured summary with key takeaways:
The Enterprise AI Revolution Is Here
We are in the early days of a major AI revolution, similar to the early cloud adoption cycle. But AI isn’t just about efficiency (like the cloud was)—it’s about transformation.
Enterprise leaders now view AI as a competitive advantage, not a “nice-to-have.”
Adoption is spreading fast, even among the most conservative sectors (e.g. finance).
AI is already reshaping job expectations and workflows, especially for the next generation of knowledge workers.
Prompts to identify where AI can drive competitive advantage:
What are the most time-consuming processes in {{department_name}} at {{company_name}}, and how might AI help streamline them?
How is {{industry}} being transformed by AI, and what’s our position in that shift?
What would it take for {{company_name}} to become an "AI-first" organization?
Who on our team can champion AI adoption in each department starting with {{priority_area}}?
If our top competitor launched an AI initiative today, how would we respond?
Intelligence Is Becoming a Commodity
“The cost of intelligence will go to zero.”
Large models (OpenAI, Claude, Mistral, LLaMA, etc.) are racing to zero marginal cost.
As foundational model costs fall, value moves to the application and integration layer.
AI capabilities will be embedded in every software product; the differentiator is how well the software solves a problem, not which model it uses.
Prompts to align business value with model abstraction:
What part of our product or service at {{company_name}} could be commoditized by AI in the next 12–24 months?
How do we make sure we’re building software that adds value on top of commodity-level intelligence?
What business processes rely on AI but do not require proprietary model development at {{company_name}}?
How are we insulating {{company_name}}’s value proposition from shifts in foundational model capabilities?
Where can we swap out models without impacting end user outcomes?
The End of the “Wrapper App” Era
“If all you built was a GPT wrapper, you probably don't have a moat.”
Simple wrappers around ChatGPT are not defensible businesses.
Instead, success will come from deeply integrated, workflow-aware, domain-specific apps.
The goal is to hide the complexity of the model and deliver a real business outcome (e.g. password reset, contract drafting, lead generation).
Prompts to differentiate beyond generic chat interfaces:
What proprietary data or workflow does {{company_name}} own that can’t be replicated by a simple AI wrapper?
How might we design a full-stack solution using AI that solves {{customer_pain_point}} end-to-end?
Which parts of our product UX would improve from task-based AI, and which require a rich GUI instead?
Are we relying too heavily on {{LLM_provider}} without building our own differentiated layer?
What could our AI do that no generic GPT interface can replicate for {{industry_segment}}?
Enterprises Don’t Care About the Model—They Care About the Outcome
Whether the underlying model is GPT-4, Claude, or Mistral matters less than whether the system:
Works reliably
Integrates with existing tools (ERP, CRM, HRIS, etc.)
Delivers measurable improvements to business processes
Prompts to align with business outcomes over technical hype:
What outcomes do our customers expect from {{product_or_service}}, regardless of which model is used?
How can we present our AI capabilities in terms of solving {{customer_goal}}, not model specs?
If we changed model providers tomorrow, how would that affect customer experience or performance?
What SLAs or guarantees can we build around outcomes powered by AI at {{company_name}}?
What workflow automation or integration is more important than which model we choose?
Intelligence Unlocks More Complex Workflows
“You can string together more agents... to complete more of the workflow.”
As model intelligence increases, we unlock more critical and complex use cases.
Think of this as a 2x2:
One axis = criticality of the workflow
Other axis = required AI intelligence
Over time, more workflows will fall within the “AI-automatable” zone, from support tickets to legal contract review to S1 filings.
Prompts to plan AI-enhanced workflows:
What sequence of tasks in {{team_name}} could be performed by AI agents in an orchestrated workflow?
How can we use LLMs or agents to move from a single-point solution to multi-step automation in {{process_name}}?
Where does current AI fail to support full workflow execution — and how do we bridge that?
What would it take to chain together {{N}} agents to fully automate {{business_process}}?
Which parts of the current workflow at {{company_name}} still require human validation and why?
Enterprises Must Understand “Core vs Context”
“Don’t build your own HR system.”
Companies need to focus AI efforts on their “core” differentiation, not general infrastructure.
Context = supporting functions like HR, CRM, payroll → buy AI-enhanced software
Core = the thing you sell (e.g. wealth management, underwriting, logistics) → build or deeply customize AI systems
Prompts to define AI investment priorities:
What is truly “core” to {{company_name}}’s value proposition, and what’s contextual infrastructure?
Should we buy or build an AI solution for {{function}}, and why?
Where can we redirect resources from building custom AI tools toward buying proven third-party solutions?
How is AI being used by competitors to enhance core offerings in {{industry}}?
If we spend {{budget_amount}} on AI, what percentage should go to context vs core?
Elasticity Creates a New Paradigm
“AI resources are elastic, unlike human labor.”
AI lets companies scale output on-demand, without needing to scale headcount.
Example: “I want 10,000 leads generated”—a task that might take months traditionally can be done in hours via AI.
This changes the structure of organizations and the business models of SaaS companies.
Prompts to redesign resource planning with AI:
What business functions could benefit from on-demand AI scale at {{company_name}}?
How do we restructure operations to take advantage of elastic AI labor for {{task_type}}?
Where are we overstaffed or under-automated that AI can help rebalance?
What is the cost per unit of output today for {{task}}, and what would it be if powered by AI?
What would a 10x scale in {{team_output}} look like using AI vs traditional resourcing?
New Business Models Emerging
“The unit of value shifts from storage or compute to intelligence delivered.”
Traditional software sold storage, compute, dashboards.
AI-first software will be sold by outcomes, work completed, or impact delivered.
This leads to new pricing models:
Outcome-based billing
Per-successful-query or task
Elastic usage of agent-based services
Prompts to rethink monetization and pricing strategy:
How could we price {{product_name}} based on outcomes or performance metrics powered by AI?
What usage-based or value-based pricing models could AI enable for {{company_name}}?
What if {{company_name}} charged by {{unit_of_measure}} (e.g. insights, decisions, resolutions) instead of licenses?
Can we build a model where AI output is the product, and human labor is optional?
What KPIs best reflect the ROI of our AI-powered features for {{customer_segment}}?
Internally: AI Will Be Everywhere
Enterprises are rolling out AI internally for:
Engineering productivity
Knowledge management (e.g. querying internal docs)
Customer service, compliance, reporting, etc.
Many are experimenting with internal chatbots, but GUIs and structured UX are still critical.
Prompts for internal rollout planning:
What internal use cases can AI automate or enhance in HR, Legal, Finance, or Ops at {{company_name}}?
Which teams are AI-ready and which need training or tooling first?
What internal LLM or RAG solutions can help staff get faster answers from {{internal_docs}}?
Should we deploy an internal assistant for {{team_name}} with access to {{knowledge_base}}?
What would a fully AI-augmented employee experience look like in {{job_role}}?
2030 Outlook: How AI Shows Up
Most AI that knowledge workers use will be embedded in third-party software (ISVs), not custom-built internal tools.
Enterprises will:
Buy AI-enhanced apps (Salesforce, Workday, etc.)
Integrate AI via APIs
Use internal orchestration layers for specific processes
Prompts to create a long-term AI vision:
What will the day-to-day work look like for a {{job_title}} at {{company_name}} in 2030?
Which of our products will be AI-native in the next 3–5 years — and what needs to change now to get there?
How should our AI roadmap evolve as foundational models continue to improve?
What partnerships do we need to future-proof {{company_name}} in an AI-first world?
If our competitor uses AI to double output, how do we stay ahead?
Final Thoughts
AI is becoming the core differentiator in enterprise competitiveness. Much like how cloud unlocked scalability, AI is unlocking intelligence and automation at scale.
The winners will be those who:
Focus on vertical and agentic workflows
Deliver outcomes, not just APIs
Balance model integration with workflow design
Understand what to build vs. what to buy