Investors should prioritize SAP and Salesforce (CRM) as durable "systems of record," as their deeply embedded business logic makes them nearly impossible for startups to replace. The most immediate actionable opportunity lies in AI startups building "usability layers" on top of these incumbents, allowing non-technical users to query complex data via natural language. Look for growth in Notion and similar "builder" platforms, which are positioned to adopt AI agents faster than traditional legacy competitors due to their tech-savvy user bases. In the industrial sector, focus on Vertical AI applications in freight and manufacturing that digitize "physical world" data exhaust and automate regulatory compliance. Avoid startups attempting to "rip and replace" core enterprise software, and instead favor "middle layer" companies that use AI to bridge communication gaps between disconnected departments like Finance and IT.

Salesforce (CRM)

• Salesforce recently announced "Headless 360," which the analysts describe primarily as a marketing rebrand of existing APIs rather than a fundamental technical shift.
• The company is attempting to position itself for an "agentic world" where AI agents, rather than humans, access CRM data.
• Salesforce remains incredibly sticky due to "inertia" and the complex business logic/customizations (SOPs) embedded in the platform over decades.
• Takeaway: While Salesforce is moving toward "headless" (UI-optional) architecture, it is unlikely to be replaced by simple database startups because it serves as an "enforcement mechanism" for sales teams and contains deep, codified business rules.

Takeaways

• Incumbent Advantage: Expect Salesforce to "bolt on" AI (e.g., AgentForce) rather than reinvent its core, relying on its status as a "system of record" to maintain market share.
• Integration Focus: Investment opportunities exist in startups that build "on top" of Salesforce to make its data more usable via natural language, rather than those trying to "rip and replace" it.

SAP (SAP)

• Described as the "ultimate example of sticky software." The analysts argue that taking SAP out of a large company like Walmart or Ford would cause the company to "evaporate" because its business rules are so deeply codified in the software.
• There is a common misconception that a modern database (like Postgres) plus APIs can replace SAP. The analysts argue this is false because the value is in the logic and customization, not just the data storage.
• SAP is currently seeing an ecosystem of startups growing around it to provide better "usability layers" (e.g., querying SAP data using natural language).

Takeaways

• Durability: SAP is viewed as nearly permanent in the enterprise due to the high cost and multi-year timeline required for implementation and customization.
• The "Usability Gap": There is a significant opportunity for AI tools that act as a "translation layer," allowing non-technical users to extract reports and analysis from SAP without needing to navigate its complex UI.

Notion

• Mentioned as a company successfully offering a "headless" product.
• Notion users are characterized as more "tech-savvy" and "agentic" (builders) compared to traditional enterprise software users, making them more likely to adopt API-first or headless workflows.

Takeaways

• User Persona Matters: Companies with a "builder" community (like Notion) may see faster adoption of AI agents than traditional legacy platforms.

Microsoft (MSFT)

• Excel and Outlook are cited as historical examples of "sticky" software.
• Excel is highlighted as a platform where users (like Goldman Sachs bankers) built their own differentiated value through macros and add-ins, creating a "network effect" within the company.
• Takeaway: The "Export to Excel" feature remains the most used "feature" in all enterprise software, representing a massive "escape valve" for data analysis that AI is now beginning to automate.

Investment Themes & Sectors

1. Headless Software & AI Agents

• Context: "Headless" means software where the User Interface (UI) is optional. In an AI-driven world, agents interact with software via APIs/code rather than clicking buttons.
• Insight: The value in software is shifting from the "head" (the UI/workflow) to the "body" (the data and the underlying business logic).

2. The "Long Tail" of Exceptions

• Context: Most enterprise work involves "exception handling"—dealing with the 20% of cases that don't fit standard rules (e.g., a custom pricing deal or a unique shipping error).
• Insight: AI’s biggest potential is in capturing and processing these "unstructured" exceptions that were previously only stored in employees' heads.

3. "Vibe Coding" vs. Enterprise Reality

• Context: There is a "wild underestimation" by startups that they can "vibe code" (quickly build via AI) replacements for complex enterprise systems.
• Risk Factor: Startups often fail to account for the massive amount of "business law," regulatory compliance (HIPAA, tax laws), and permissioning required in large-scale software.

4. Vertical AI (Physical World)

• Context: Sectors like construction, manufacturing, and freight are mentioned as high-opportunity areas.
• Insight: These industries have "physical realities" and "data exhaust" that have been hard to capture historically. AI agents (like voice agents for freight compliance) are now starting to digitize these offline interactions.

5. The "Middle Layer" Opportunity

• Context: The best opportunity for new startups is to sit between two established legacy players (e.g., between Finance and IT).
• Insight: Legacy vendors are unlikely to disturb their own "go-to-market" strategies. Startups that bridge different functions using AI to help them communicate can create entirely new software categories.