We set about understanding industry specific business challenges with the goal of accelerating the time it takes to bring Applied AI to regulated and complex markets.
The Importance of Deeply Understanding Business Requirements
In the last few months, you might have noticed we've not been at every AI event. Instead we shifted our focus away from engaging with technologists, and instead we decided to go deep with industry experts in Healthcare, Oil and Gas, and more. The primary goal of this shift is to deeply understand their business requirements and learn what is really needed by them. By understanding their needs, we evolved a new approach to deliver Applied AI into production.
All things being eventually equal - resources, skills, tooling, capability, and capacity - ultimately the only competitive advantage is time. Applied AI systems compound in value, and first movers can establish a significant advantage.
An important lesson is that no executive cares about the model that runs underneath the outcome. What matters to them is getting repeatable, manageable, and adaptable results. I've yet to meet one that hasn't echoed this to our team.
We have also realized that no one needs yet another standalone app. Industries have incredible app sprawl, which most leaders consider an exposure, but a necessary compromise. Applied AI needs to be invisible to the end user. It should seamlessly integrate into their existing workflows and systems. Humans in these industries still hold the experience, knowledge, and logic that is and will continue to be immensely valuable for the foreseeable future.
When defining problems, it is essential to consider the pain, impact, exposures, and mitigations. By clearly understanding the problem, we can develop effective solutions. Most executives can estimate the impact of a problem immediately when we ask, and can also recite what has been spend on the problem to date. Very few technologists are able to do this with the same rigor.
While chatbots are cool and understandable, they are not the only agents that drive impact. There are many other other avenues to explore for maximizing the impact of Applied AI. Today, Data Kinetic implements chat agents for the last mile, but the real magic happens upstream for that through chain of thought, search, and many others.
Lastly, we must navigate the market carefully. Many vendors are flooding the market with confusion and misplaced promises on capability. In some cases, they are simply AI washing their products. It's crucial to avoid falling into these traps. To my friends in marketing, my advice is be honest, clear, and crisp. Industry executives want clear, concise answers, and you gain more credibility from explaining what you can't do vs what you can do.
Key Lessons Learned from Industry Experts in Healthcare, Oil and Gas, and More
During our deep dive with industry experts in Healthcare, Oil and Gas, and other sectors, we have uncovered a number of lessons. I want to thank many of our new friends in Healthcare, Oil and Gas, Insurance, FinServ, who were patient with us, and took the time to share their real-world problems. Their expertise has been instrumental in developing a new approach that meets the unique needs of each industry.
Delivering Applied AI into Production: Developing a New Approach
Our foremost objective is to bring Applied AI into operational use. In pursuit of this, we've crafted a new methodology, drawing from the wisdom, scar tissue, and experiences of industry professionals. This strategy is designed to thoroughly comprehend our clients' business needs and take advantage of system blueprints that align with those requirements.
This also led us to develop a number of new toolsets and technologies to enable this. More on those in the near future. However, I will say that by rethinking how we defined the problem, it totally changed the way we thought about how we build outcomes for industry.
By focusing on developing a new approach, we can overcome the challenges and complexities associated with implementing Applied AI. This approach enables us to deliver results that are repeatable, manageable, and adaptable.
Creating an Invisible User Experience: Moving Away from Standalone Apps
Our primary mission in delivering Applied AI extends beyond just technological integration; it emphasizes the crucial role of 'human over the loop' in the industry. Operators possess a wealth of deep knowledge, logic, and experience that isn't always directly captured as data. Instead, these valuable insights are often implicit within the data. Recognizing this, our approach to Applied AI is not just about processing data but also about understanding and leveraging the human expertise embedded within it.
One pivotal aspect of our strategy in deploying Applied AI is to create an 'invisible' change to the user experience. We've come to understand that the market does not need another isolated application. Rather, Applied AI should be intricately woven into the existing workflows and systems of the end users. This seamless integration is key to making the technology an intuitive and unobtrusive part of daily operations.
By steering clear of standalone applications, we ensure that Applied AI is not just an add-on but a fundamental component of the system and user experience. This integration approach is crucial for enhancing usability, and capturing the implicit knowledge within organizations. It allows users to benefit from advanced AI capabilities without the need to adapt to a yet another new tool or interface. The technology works in harmony with their existing systems, complementing and enhancing their work rather than disrupting it.
This approach respects and utilizes the deep-seated expertise of industry operators. By integrating AI into their familiar environments, we enable these professionals to apply their insights more effectively, using AI as a tool that amplifies their knowledge and experience rather than replacing it. This synergy between human expertise and AI technology is what truly unlocks the potential of Applied AI, making it a powerful asset in any industry setting.
Defining Problems Effectively: The Importance of Pain, Impact, Exposures, and Mitigations
Expanding on the importance of defining problems effectively in the development of successful solutions, it's essential to consider not only the immediate pain points and impacts but also the broader context in which these problems exist. This includes both internal dynamics, such as team silos within an organization, and external forces, like regulatory constraints.
When we talk about internal factors like team silos, we're referring to the challenges that arise from segmented or isolated groups within an organization. These silos can lead to a lack of communication and collaboration, potentially issues with resource allocation, often resulting in a fragmented understanding of problems. By recognizing these internal barriers, we can tailor solutions that not only address the technical aspects of a problem but also facilitate better integration and cooperation among different teams. This holistic approach ensures that solutions are not developed in isolation but are instead aligned with the overall organizational structure and culture. Together we can usually move faster, focus on specific measurable outcomes, and avoid over-scoping and creating never-ending projects usually die or get pivoted.
External forces, such as regulatory requirements, also play a crucial role in problem definition. Regulations often dictate the parameters within which solutions must operate, especially in industries like finance, healthcare, and telecommunications. Understanding these regulatory environments is crucial for developing solutions that are not only effective but also compliant. This involves staying abreast of current and upcoming regulations and integrating this understanding into the problem-solving process. In many cases, we've literally read the regulations directly (or had a model do it to then have experts approve these rulesets), and then in turn made those part of the models themselves.
Beyond Chatbots: Exploring Other Agents of Impact
While chatbots are often seen as the face of Applied AI, they are not the only agents that drive impact. There are other avenues to explore for maximizing the impact of Applied AI in various industries.
Not all agents need to be masters of the universe, rather, single task, single focus, highly controlled, and highly tuned agents and models are in our experience far more likely to generate a predictable result, vs a huge model being asked to solve all problems. The teams at Meta, and OpenAI are phenomenal, but just because these models can create large diverse outcomes, it doesn't mean that in Applied AI we should be necessarily trying to do the same. In fact in many cases, it might be doing more harm that good to try and leverage LLM's in this fashion for complex enterprise use cases.
Navigating the Market: Avoiding Confusion and Misplaced Promises
The market is flooded with tech vendors offering AI solutions, but not all promises are genuine. Many vendors create confusion and make misplaced promises about their capabilities.
To navigate the market successfully, it is crucial to critically evaluate vendors and their offerings. We need to be cautious of AI washing, where vendors exaggerate the AI capabilities of their products. By avoiding these traps, we can make informed decisions and choose the right partners for our AI initiatives. Remember, no executive really cares how cool your AI is, they're trying to positively solve a problem, not marvel at how many GPUs or parameters you're working with. My advice is stay focused on a specific outcome or project when making decisions, but during the decision process also consider your next two projects. Remember good Applied AI drives compounding outcomes - even specialized models, can build off each other if architected correctly.
What next?
Going forward we're going to spend time sharing more about what we've been working on, how we see it supporting the industries we're focused on, and ultimately how we can continue to contribute to the broader ecosystem at large. Over the next few week's I'll be sharing some of the trends we're seeing emerging, and diving deeper into how a different approach to Applied AI might be necessary for all of us going forward.
