The AI Canvas: A methodology for successful AI transformations
The goal is clear: creating value in your company with AI. The way to get there is bumpy and challenging, though. As a holistic tool, the AI Canvas offers clear guard rails and steps that lead to success in the implementation and scaling of AI projects.
AI adoption is lagging behind its potential
In recent years, research in the field of artificial intelligence (AI) has made significant progress. It has also become quite clear that companies can gain long-term competitive advantage if they successfully employ AI in their operations. However, according to a CEO study, only 65% of the companies surveyed have a comprehensive AI strategy (Deloitte, 2019). It seems that there is still a large discrepancy between the potential and implementation of AI.
Looking at companies in terms of their AI maturity, there are several types; while many have recognized the added value of the technology, for some, the full potential of AI is still untapped:
Why is AI implementation lagging behind?
The reasons why companies find it challenging to implement AI are complex but can be traced back to the following points:
- Choosing the right business case: Most companies understand AI as a means to an end to make processes more efficient or to automate repetitive decision-making tasks. The expectation of a fast ROI is therefore usually extremely high. However, companies with a lack of experience are often unable to assess which dynamics are important for data-driven business models in the medium term. Therefore, most of them find it difficult to identify the strongest cases and to bundle their resources accordingly.
- Choosing the right organizational setup: From an organizational point of view, there is often a lack of synchronization between decentralized initiatives (bottom-up) and the overarching corporate strategy (top-down). Establishing processes and structures for a corresponding alignment is an important but difficult step in the process of AI implementation.
- Understanding the technology in a holistic manner: When developing and deploying AI models, it is not just the algorithms that need to be considered and thought through. Strong attention should be paid to topics such as data quality, security, or regulatory issues. Unfortunately, organizations often lack expertise in these topics which can pose a future risk.
The need for a comprehensive AI methodology
In order to profitably exploit the possibilities of AI, strategic planning and organizational flexibility must come together. A holistic methodology that aligns fast-running processes with long-term goals is necessary for the successful implementation and scaling of AI projects.
AI projects are characterized by their dynamic nature and inherent uncertainty. Internal and external requirements, changes in data or the technological infrastructure have major and often unpredictable effects on the success of a project. It is therefore necessary to develop an understanding of all of these factors and incorporate them into high-level strategic planning. Dataset drifting, changing external requirements and the rapid development of AI require an iterative approach and monitoring of AI products. In contrast to conventional technologies, AI affects almost every part of an organization. The creation of transparency and understanding is just as necessary as a coordinated, holistic approach to the implementation of AI.
The AI Canvas: A strategic toolkit for your AI initiatives
The AI Canvas, a strategic toolkit for the implementation of AI projects, provides an answer to the need for a strong methodology fully tailored to the topic of AI.
The AI Canvas can be used across the whole process of AI lifecycle — from ideation, through software building, implementation and scaling, to long term AI strategy development. The methodology is specially tailored to AI projects and their characteristics. As a result of joint work of Merantix Labs and University of St. Gallen, the AI Canvas combines the latest findings from science and business applications. The hierarchically structured framework addresses the described complexity drivers of AI projects and is divided into four pillars of business, organization, technology and lifecycle at the top level (www.merantixlabs.com/ai-canvas).
Each of the blocks has an essential role in AI initiative planning and implementation:
- BUSINESS: Understanding problems and domains, as well as translating requirements into AI-based products / services, and converting them into measurable performance indicators, are essential components for the evaluation of AI investment projects.
- ORGANIZATION: It is crucial to set up a company in a way that it can use resources and processes to drive rapid innovation cycles. Understanding the inner workings and talents of a given company helps in building the most effective ecosystem possible for the development and scaling of AI.
- TECHNOLOGY: In addition to questions about data and technological setup, the novelty of many AI applications creates challenges at the operational level. Regulatory, liability and safety-related aspects become important when AI applications grow more complex.
- AI LIFECYCLE: A key characteristic of AI systems is their dynamic nature. Changes to the underlying data distribution, the KPIs or the organizational set-up require a re-evaluation of the technology and the models. AI products go through a life cycle that includes all pillars (business, organization, technology) of the AI Canvas.
Depending on the AI maturity level of an organization, it can be used at the use case, scaling or even organizational level.
As an example, we partnered up with a large publishing house to develop a pilot and set up a long-term AI strategy. Using the AI Canvas at the beginning of the engagement has proven extremely useful. We gathered leaders from different departments who don’t usually collaborate and discussed how to scale an already existing AI pilot and turn the company into a data-driven organization
While using the AI Canvas, we found out that the existing organizational structure resulted in siloed functioning of various business lines which were not effectively sharing resources or communicating on the topic of AI. Together we developed a functioning AI hub setup ensuring that knowledge across siloed business lines was shared, which initiated the creation of an AI ecosystem. The resulting AI hub has since contributed to sustainable talent acquisition for the interdisciplinary project teams required as part of the AI strategy. The AI hub ensures that individual project ideas contribute to strategic priorities and that different departments from the company are coordinated and synergies are used efficiently.
Some of the business lines also worked with highly sensitive data. With the help of the AI Canvas, these findings were systematically transferred into requirements for the data infrastructure and the technological setup. The requirements were implemented operationally through a sharp separation of individual database schemes, the unification of standards and the redefinition of interfaces. Synergies between individual business lines with regard to data and infrastructure quickly became transparent.
Using the AI Canvas proved to be an effective tool for establishing a permanent setup for future AI projects. Today, the identified synergies (talent acquisition, data & infrastructure) make a significant contribution to the transformation into an AI-driven business model.