# Transitioning from AI-Assisted to AI-Native Workflows Key strategies for integrating AI into organizational structures and workflows. - **Source:** [O'Reilly Radar](https://www.oreilly.com/radar) - **Published:** 2026-09-18 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/6ad28c04-a31c-4501-8543-7fa154d0c0d1 ## The bitter lesson of AI Richard Sutton's 'bitter lesson' highlights that brute force and compute scale often surpass human expertise in AI development. This principle has been evident in tech history, from Google's dominance over manually curated directories like Yahoo to ChatGPT's recent impact. Companies must recognize this dynamic to avoid being outmatched by competitors leveraging AI more effectively. The lesson underscores the need for organizations to adapt their strategies around AI's capabilities rather than clinging to traditional methods. ## AI-native mindset shift Transitioning from AI-assisted to AI-native workflows requires a fundamental mindset change. Initially, AI might be used for tasks like generating quiz questions, but true integration involves reimagining products to reflect current AI capabilities. For example, O'Reilly evolved its skill path builder into an interactive experience by leveraging LLMs to identify developer skills. This shift demands a reconceptualization of workflows to fully harness AI's potential, moving beyond superficial assistance to deep structural integration. ## Human challenges in AI adoption AI adoption is as much a psychological challenge as a technical one. Trail of Bits CEO Dan Guido found that only 5% of his staff was initially on board with AI transformation, while 70% were passive and 20% resistant. Key barriers include self-enhancing bias, opacity, intolerance for imperfection, and identity threat—the fear that AI might replace not just tasks but professional identities. Overcoming these requires framing AI as a tool that enhances, rather than diminishes, human roles. ## Capability ladders for AI integration Trail of Bits uses a three-level status ladder to guide AI adoption: Level 0 (resisting AI), Level 1 (experimenting with AI), and Level 2 (building AI to enhance organizational capability). Each level is tied to specific departmental matrices, ensuring relevance across roles. For instance, security auditors and accountants have distinct AI applications. This structured approach helps employees understand their current stage and focus areas, aligning individual progress with company-wide AI goals. ## Hackathons as knowledge accelerators Trail of Bits runs bi-monthly hackathons with predefined objectives and learning goals to foster AI adoption. Success is measured by participants' advancement on the capability ladder, not just deliverables. Outcomes are documented in a shared skill repository, creating reusable artifacts for the entire company. This iterative process ensures continuous learning, with each hackathon building on previous insights, compounding expertise across the organization. ## Standardizing AI safety and infrastructure To avoid 'prompt debt'—overly complex, model-specific prompts—Trail of Bits maintains a company-wide repository of fixes and standardized safety nets. Sandboxes cater to diverse needs, and a seven-day cooldown period is enforced for new external packages. These measures ensure consistency and reduce friction in AI adoption. CEO Dan Guido emphasizes the importance of unstructured experimentation time, a lesson learned from initial implementation gaps. ## Human expertise as a differentiator While AI enhances productivity, it doesn't replace human expertise. O'Reilly's Expert Intelligence initiative leverages AI to deliver practitioner-authenticated insights, such as workload analyses and hiring cases. This grounding layer integrates AI with human knowledge, ensuring decisions are informed by real-world expertise. The approach highlights AI's role as a medium for extending, not replacing, unique human perspectives and skills. ## Strategic implications for organizations The transition to AI-native workflows demands more than technical upgrades—it requires cultural and structural shifts. Companies must address psychological barriers, standardize processes, and foster continuous learning through tools like capability ladders and hackathons. Success hinges on viewing AI as a complement to human expertise, enabling teams to tackle larger challenges. As Tim O'Reilly notes, the goal is to 'celebrate AI's small wins and focus on bigger problems with newfound capabilities.' ## Sources covering this story - https://www.oreilly.com/radar/how-to-get-from-ai-assisted-to-ai-native/ — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._