# Atlassian CEO on the SaaS Apocalypse, AI Agents & What Comes Next Source: https://www.youtube.com/watch?v=0lzo2tFBFy8 Recap page: https://rapidrecap.app/video/0lzo2tFBFy8 Generated: 2026-03-06T16:03:52.625+00:00 --- ## Quick Overview The main outcome is that while AI tools like those from OpenAI offer significant power for automating tasks, businesses must fundamentally rethink their processes, moving away from industrial-era, input-constrained thinking toward outcome-based, auditable, and transparent workflows to fully realize AI's value and avoid pitfalls like over-reliance on non-deterministic outputs. **Key Points:** - The capabilities of current AI models far exceed the value they are currently delivering to businesses, often relying on antiquated, input-constrained process models. - The speaker contrasts the industrial-era model (where processes are fixed, like a filing cabinet) with the AI-era model, which requires focusing on outcomes and transparency. - Companies like Salesforce are moving toward outcome-based pricing, but the underlying operational processes (like those at GE from the 1960s) often remain rigid, leading to inefficiencies. - The speaker champions AI tools that allow for iterative development and transparency, contrasting this with relying on opaque, complex workflows. - The ultimate goal is to shift from highly constrained, manual processes (like manually retrieving files or HR processes) to flexible systems where AI can be used to iterate quickly on desired outcomes. - The speaker notes that while AI excels at tasks like document creation, the challenge lies in building a fundamental platform that supports these new, dynamic workflows, rather than just patching old systems. ![Screenshot at 00:05: The screen displays a visual transition showing code changes \(model: 'gpt-5-nano' being replaced by 'gpt-4.1-mini'\), illustrating the rapid iteration and model evolution central to the AI discussion.](https://ss.rapidrecap.app/screens/0lzo2tFBFy8/00-00-05.jpg) **Context:** This video features a discussion, likely an interview or podcast segment from the a16z Show, involving at least three participants discussing the impact of advanced AI tools on business processes, particularly within the context of SaaS companies. The conversation centers on overcoming outdated, rigid systems designed for the industrial era (like fixed workflows and manual processes) and adapting to the new capabilities offered by AI, emphasizing the need for transparency and outcome-based thinking in system design and pricing models. ## Detailed Analysis The discussion argues that the current capabilities of AI far outstrip the value businesses extract because they attempt to fit powerful AI tools into outdated, industrial-era process frameworks that are input-constrained (like filing cabinets or rigid workflows). The speaker contrasts this with the potential of AI to enable outcome-based, transparent processes. An analogy is drawn to the history of software development since 1960, where moving from physical filing systems to databases, while beneficial, didn't fundamentally change the underlying human dependency for querying information. The speaker cites examples like Salesforce's eventual shift to outcome-based pricing, suggesting that pricing should reflect the value delivered (e.g., 10x output for the same input cost) rather than seat licenses. The core challenge is that many companies are still trying to solve complex, iterative design problems (like improving customer service or compliance) using deterministic, rigid systems. The speaker advocates for a fundamental shift in design, where AI tools are integrated to allow for rapid iteration on desired outputs, rather than simply automating existing, inefficient processes. The speaker highlights that while AI can perform tasks like writing documents or summarizing tickets, the underlying organizational structure and trust mechanisms (like how agents are assigned work) must evolve to leverage this new power effectively. ### AI Adoption & Process Rigidity - AI models are far ahead of current business processes - Models are being constrained by input-focused thinking rather than outcome-focused needs - Many companies still operate on industrial-era models that hinder AI adoption. ### Pricing & Value Models - Outcome-based pricing (like per-use) is replacing seat-based licensing - The value delivered by AI tools (e.g., 10x efficiency) should be reflected in pricing, making per-seat models unfair. ### The Iteration Challenge - AI excels at generating initial drafts (like essays or code) but the iterative refinement process is manual and time-consuming - This highlights the need for AI tools that support continuous improvement cycles, not just one-shot generation. ### Fundamental Design Shift - The goal is to move from rigid, pre-defined workflows (like traditional HR or compliance systems) to flexible, dynamically driven systems - This requires building new foundational platforms rather than just patching old software. ![Screenshot at 00:05: The screen displays a visual transition showing code changes \(model: 'gpt-5-nano' being replaced by 'gpt-4.1-mini'\), illustrating the rapid iteration and model evolution central to the AI discussion.](https://ss.rapidrecap.app/screens/0lzo2tFBFy8/00-00-05.jpg) ![Screenshot at 00:15: A graphic illustrating data filing cabinets transforming into abstract, colorful vertical bars, symbolizing the shift from legacy data storage to modern, flexible systems.](https://ss.rapidrecap.app/screens/0lzo2tFBFy8/00-00-15.jpg) ![Screenshot at 00:20: Text overlay highlighting key metrics of an AI assistant: "It is doing the equivalent work of 700 full-time agents" and achieving human-level customer satisfaction scores.](https://ss.rapidrecap.app/screens/0lzo2tFBFy8/00-00-20.jpg) ![Screenshot at 02:22: A visual demonstration of AI assisting in coding, showing a request to "Add unit tests to the checkout flow to cover payment failure scenarios" within a code editor interface.](https://ss.rapidrecap.app/screens/0lzo2tFBFy8/00-02-22.jpg)