Is AI for All the Answer to Your Business Problem?

“AI for all of us” is the tagline for Claude, a cloud native large language model (LLM) offering from Anthropic. The slogan goes on to say that whether brainstorming alone or working with a team of thousands, Claude is the tool to help you achieve your results. However, the assertion that a model is “AI for all” comes with some questions that privacy and security minded business and government leaders should be asking. Do you want AI that is for all? Or do you want AI that is for you? The rise of large, public cloud based LLMs has led to impressive innovations in AI technology and large investments in innovators around the world. Social media is full of examples of images, sonnets, and blogs that were all LLM generated. These posts are fun and, in many cases, have saved considerable time for their creators. Some of them are awful and make you wonder if it’s killing art. But there are other stories about LLM failures such as the New York lawyer who was sanctioned for using ChatGPT to create a legal briefing that included non-existent precedent cases or the Chevrolet chatbot that sold a 2024 Tahoe for $1. The lesson? Maybe general purpose LLMs, “AI for all of us,” are not the answer. Maybe what we need are customized LLMs that are specifically trained and tuned for the specific use case vertical for which they are intended. The technology exists today to adopt fully secure, privacy preserving AI that is trained only on the data that is relevant to its operation. This reduces the risk of inaccurate results and creates specialization in AI during a time when big AI companies are creating generalist applications. So now I ask you: Do you need a specialist or a generalist? AI Access There is a significant difference between “AI for all” and democratizing access to AI. One of the areas for which the big AI companies deserve credit is in building tools that could be accessed by all. Many people have learned about AI simply from the ability to actually use one rather than read about AI in the abstract. Democratizing the ability to use AI, understand its operation, and understand its training processes is a goal that we should share across society. AI is, and will continue, playing a major role in how we work, how we support our customers, and how we augment our human domain experts. Access to AI should be for all. But as a business leader, should your internal AI be for all? Share AI for all implies a generalization of the tools and algorithms in question. It also implies a generalization of the data used to train it. That works well for the democratization of AI mentioned above but it is not the answer for business and government leaders who are trying to create value or solve a mission challenge on a specific issue. This generalization becomes a barrier to entry for many organizations as they have a difficult time evaluating the risks and understanding the value proposition. An algorithm whose training includes the terabytes of data on cetacean behavior in the Southern Ocean is not going to help a hedge fund portfolio manager trying to generate alpha any more than it will help a Security Operations Center operator responding to a cyber incident. This means that we need to stop thinking in terms of generalized AI tools when what we need is a specialized tool for a specialized problem. Here are some things to keep in mind: Risk : Understanding the risk to a mission or organization from the use of any AI tool is difficult to impossible when the purchasing organization has little to no visibility into the training process and dataset. Leaders should know what data has trained their algorithm from its inception so they can build a true risk picture. Value : Whether serving customers or responding to a national security incident, leaders should have an expectation of exactly how much value they can extract from their AI implementation. Generalist AI makes this question close to impossible as there are too many unknowns such as the training dataset, hallucination risks, and deterministic accuracy. Conversely, customized specialist AI allows leaders to see the expected value of their AI deployment and to create metrics and measures of success against those expectations. This smooths the path to implementation by giving realistic measurements that show a clear ROI. Subscribe now Transparency : The hit to the reputation of an organization following a cyber or AI incident can be truly catastrophic to its bottom line. This risk keeps many executives from implementing AI when it could otherwise create value. Transparency in AI is difficult to come by with the large AI companies such as Anthropic, Google, and OpenAI; but this does not have to be the case. An organization implementing AI should have full transparency into the training, tuning, operation, and output of their system. Leaders should seek AI partners that provide them the opportunity to quality check the output of the system to give an additional layer of transparency to the process. This gives organizations internal and external transparency that will turn into institutional value alongside direct value from the use of AI. What’s Next Democratizing AI has been, and will continue to be, a significant benefit across industries and sectors. Having access to actually use AI tools will broadly increase understanding of AI and will lead to new innovations and use cases. But for leaders of organizations considering using AI for value generation, “AI for all” means something very different. AI for all implies generalization, which is not going to drive the most value for the implementor. Generalist AI also limits transparency, harms risk analysis, and puts up internal barriers to entry. Specialist AI is what’s next. Specialist AI allows organizations to understand exactly what data the algorithm was trained on and what actions were taken to tune the model. It also creates a truly proprietary AI system that knows your specific use case extremely well and is not distracted or confused by irrelevant data. This results in higher quality outputs, which result in more value. Wealth as Smith would say. For leaders considering AI for their specific use cases in 2025, specialization is how to find value and mitigate risks. Democratize AI in your workforce through the “AI for all” tools but produce value from AI for you. Connect with us: LinkedIn , Bluesky , X , Website To learn more about the services we offer, please visit our product page. Nick Reese is the cofounder and COO of Frontier Foundry and an adjunct professor of emerging technology at NYU. He is a veteran and a former US government policymaker on cyber and technology issues. Visit his LinkedIn here . This post was edited by Thomas Morin, Marketing Analyst at Frontier Foundry. View his Substack here and his LinkedIn here . Leave a comment