Ernest Hemingway, The Sun Also Rises
In our previous blog post, "AI Reshaping Companies: Lessons for Us All," we explored how Klarna the global Swedish BNPL company created worldwide headlines and attention about the revolutionary reshaping of their business with AI. Using AI, they had eliminated costly software-as-a-service (SaaS) solutions, reduced staffing needs, and dramatically improved personalisation in customer services. Klarna achieved this transformation by phasing out key SaaS tools, starting with Salesforce and then removing Workday. While their journey exemplified the anticipated impact of AI - the move from Software as a Service (SaaS) to Service as a Software - it was the first significant realisation that the warnings were real.
In this blog, we explain the emergence of "service-as-a-software" and its implications for enterprise AI. We outline the shift from service-dominant workflows to software-dominant workflows, introducing the O33 AI Framework that makes this transformation possible.
The shift to AI-driven processes involves a cycle of unbundling and rebundling tasks, transforming human-dependent services into software-dominant workflows in the following process:
1. Service-Dominant Workflow: Initially, workflows are largely dependent on human decision-making and actions. Software plays a secondary role, managing simpler tasks like data processing or basic automation. Human workers are responsible for the majority of decision-making and problem-solving, making workflows slower and more prone to human error. The reliance on human judgment means that scaling these workflows can be challenging and costly.
2. Unbundling: As AI capabilities advance, it becomes possible to take over more complex tasks. Organisations can identify parts of workflows where AI can replace or enhance human effort, gradually reducing manual tasks. AI systems are deployed to analyse workflows, pinpointing specific areas where automation can lead to increased efficiency. By breaking down these workflows, companies can see which tasks are best suited for AI integration, allowing human workers to focus on more strategic and creative responsibilities.
3. Componentising (Service-as-a-Software): AI enables the componentisation of specific tasks into independent software modules. These modules can remain within their original workflows or be integrated into other processes via APIs, allowing for greater flexibility and reuse. Componentising tasks into software modules means that organisations can create a library of reusable AI-driven components. These modules can be easily adapted and plugged into various workflows, enabling a higher degree of customisation and interoperability between different systems.
4. Rebundling: These software modules are then reorganised into new or optimised workflows, to improve efficiency and adapt to organisational needs. However, rebundling also provides an opportunity to rethink the entire process. Instead of merely replacing human tasks with software, businesses can redesign workflows to maximise efficiency, eliminate redundancies, and improve outcomes.
5. Software-Dominant Workflow: After rebundling, we reach a stage where most tasks are executed by software, with minimal human involvement. This process is iterative - as AI technologies progress, workflows need to be continuously refined to absorb more tasks into software. In a software-dominant workflow, human workers are still present but focus primarily on overseeing AI systems, addressing exceptions, and handling tasks that require uniquely human skills like empathy or complex problem-solving.
The journey of unbundling and rebundling starts with first principles: work is a bundle of tasks aimed at achieving specific goals.
Tasks can be performed by humans or software, depending on their complexity. While most routine services have already been automated, two categories of work still require human intervention:
Generative AI has begun to shift this dynamic by breaking work into individual tasks previously carried out by humans and rebuilding them as software components. The transition from service-dominant workflows to software-dominant workflows occurs as AI increasingly takes on both knowledge and managerial tasks. AI's ability to process vast amounts of data, recognise patterns, and generate insights means that it can take over tasks that were once considered too complex for automation. This allows businesses to operate more efficiently and with greater precision.
The success of this shift relies on several key factors:
The O33 AI Framework has been designed to facilitate this transition. The framework offers agentic, multiagent capabilities derived from AutoGen (Microsoft Research) and Microsoft’s Semantic Kernel. The architecture enables tasks to be broken down into independent software modules that specialised execution agents can handle, all orchestrated by management agents responsible for the overall goal.
The framework's multiagent architecture allows for the delegation of tasks to specialised agents that operate autonomously. This enables a high degree of parallelism, where multiple agents work simultaneously on different aspects of a workflow, significantly speeding up processes. The management agent oversees the coordination, ensuring that all tasks align with the broader objectives and that any issues are addressed promptly.
Unlike non-agentic AI, such as chatbots and copilots that require direct human input, agentic AI systems function autonomously and adapt to changing situations without constant human oversight. While many other providers only offer non-agentic solutions like chatbots and copilots, the O33 AI Framework provides the full spectrum of agentic capabilities. This means that not only can you customise, modify, and build non-agentic chatbots and copilots, but you also gain access to sophisticated agentic AI systems that operate with a higher degree of autonomy. This adaptability is crucial for moving away from service-dominant workflows and toward software-dominant, efficient, and scalable operations. Agentic AI can learn from experience, adjust its strategies in real time, and handle unexpected challenges, making it ideal for complex enterprise environments where conditions can change rapidly.
The O33 AI Framework is flexible - it allows modules as well as execution agents to be swapped-out or upgraded as new and improved large language models are developed. This ensures that organisations can continue to innovate and remain competitive as technology evolves. By maintaining a componentised and modular structure, the framework allows for easy integration of new capabilities, ensuring that businesses can take advantage of the latest advancements in AI without significant disruptions to their operations.
Generative AI is not just about automating tasks—it’s about reimagining how businesses operate by unbundling human services and rebuilding them as efficient, scalable software. This transformation requires a shift in mindset, moving from seeing AI as a tool to viewing it as an integral part of the service delivery model. Are you ready to embrace this shift? Reach out to us to learn how the O33 AI Framework can help you unlock the potential of AI for your enterprise.
Generative AI offers a path to unprecedented efficiency, scalability, and innovation. By leveraging the power of AI to unbundle and rebundle tasks, organisations can build smarter workflows, improve customer experiences, and gain a significant competitive edge. The future is not just about doing the same work faster - it’s about redefining the work itself. Let us help you navigate this exciting transformation and position your business for success in the AI-driven era.
Transform the way you operate with our advanced AI Framework and Semantic Kernel solutions.