One thing I've struggled with is separating in my mind the different ways AI tools are useful to individuals versus how they are useful for businesses and organizations, and how these uses interact with each other. For individuals, AI models are best used to expand your own personal capabilities. They can help you become a better data analyst, a more creative brainstormer, or stand in for a knowledgeable expert opinion. They can analyze contracts for red flags, provide quick first drafts, or offer summaries of complex documents tailored to your specific needs. It takes working with the tools for a while to truly understand what they're good at and how they can help you improve your work. Improvement in your skills can happen very quickly once you become comfortable using the tools. You'll learn the most by applying different AI tools to as many problems as possible, iterating on prompts to discover the strengths and weaknesses of the models. On the business side, AI models can be integrated into workflows, often by bringing the organization's data to the models for analysis, transformation, or generating insights that can automate business processes or inform decision-makers. Significant improvements and efficiencies for businesses may take longer to realize because workflows can be fragile and complex. Organizations won't have the ability to quickly iterate like individuals to find what works. Developing AI-enhanced workflows might require expensive new subscriptions, specialized staff, or new skills for existing staff—all challenges that organizations may struggle to get right. Although I'm not going into detail on this, data security, confidentiality, and compliance issues will also have to be addressed before starting these kinds of projects. No doubt some of these challenges will prevent certain organizations from pursuing these kinds of large-scale projects. They will have to rely on unlocking efficiencies through the incremental improvements from employees' personal use. These organizations will need to encourage their staff to experiment with AI tools, identify individuals who have found clever, useful ways to use AI to speed up workflows or solve problems more efficiently, and try to disseminate those ideas across the organization. I suspect these methods of using AI, pioneered by individuals, will be good candidates for organizations to consider when developing or integrating AI models into their workflows.