Why AI Initiatives Aren't Just Projects

And that's the reason why we need AI Product Managers - Recognizing the True Scope and Effort Behind Successful AI Initiatives.

#beyondAI

"Calling AI initiatives mere projects is like trying to build a skyscraper with the mindset of putting up a tent." 🏢✨ This is a message I often convey to those embarking on their AI journey.

If companies continue to view AI initiatives through the narrow lens of projects, they risk falling into a cycle of unmet expectations and failures. This misguided perspective can lead to resource misallocation, stagnant innovation, and ultimately, the inability to leverage AI for sustainable growth and competitive advantage.

In this issue, we will delve into why AI initiatives require a broader, more dynamic approach. We’ll explore the multifaceted nature of AI efforts and highlight the necessity of treating these initiatives as evolving products rather than finite projects.

You will learn about:

• The Complexity of AI Initiatives: The Double Trio

• Dynamic Nature of the Double Trio

• Choosing Between Project and Product Approaches

• Final Thoughts on Embracing a Product Mindset

Happy reading 🛋️

---

---

The Complexity of AI Initiatives: The Double Trio

Companies still complain that their AI projects fail, and the reasons are already clear. However, there is a misconception, or even a romantic imagination, that once deployed, AI will deliver continuous value and funding can be stopped. I believe AI initiatives in companies should not be approached with a project mindset but rather a product mindset.

AI in companies does not behave like simple automation tasks, where there is a clear, non-dynamic input and a well-formed output can be expected. If you ever find such a case, know that it is a simple rule-based automation system and not one that requires AI. You can read here how to decide when you need AI, and when it is appropriate to go with No-AI.

If companies want to implement AI into their internal processes or augment their labor capability with an AI assistant, they need to understand that the complexity of their initiatives is defined by six dimensions:

The Technical Trio:

• Data

• IT

• AI

The Strategic Trio:

• Business

• Governance

• People

I call them the Double Trio.

Unfortunately, every single element of the Double Trio behaves dynamically, some more so than others. Let’s understand why we deal with such a dynamic environment when we want to apply AI in companies:

The Dynamic Nature of …

... Data

Data is the lifeblood of AI. The dynamic nature of data means that sources, quality, and relevance are constantly changing. Collection methods evolve, new sources emerge, and the volume increases exponentially. Additionally, data privacy regulations frequently change, necessitating continuous updates to data handling practices. This volatility requires ongoing monitoring, management, and adaptation to maintain data integrity and usefulness in AI models.

… IT

IT infrastructure, the backbone of any AI initiative, is inherently dynamic. Advances in hardware, cloud computing, and software development continuously reshape the landscape. Integration of new technologies and updates to existing systems are ongoing challenges. Furthermore, IT governance policies and security protocols need constant revision to address new threats and compliance requirements.

… AI

AI technology evolves rapidly. New algorithms, frameworks, and tools are regularly developed, offering improved capabilities. AI models require regular tuning, retraining, and updating to maintain accuracy. As data distributions change, models can become outdated, necessitating continuous improvement. Significant research and development in AI lead to frequent introduction of new techniques that enhance existing solutions.

… Business

Business environments are highly dynamic, influenced by market trends, competitive pressures, and internal strategic shifts. Objectives and requirements can change quickly, impacting AI initiatives. Companies must align their AI projects with evolving goals to deliver value. Changes in customer behavior, regulatory landscapes, and economic conditions also affect strategies, requiring AI solutions to be flexible and adaptable.

… Governance

Governance encompasses policies, regulations, and ethical considerations. Regulatory landscapes continuously change, with new laws addressing emerging AI issues. Compliance with data protection regulations, ethical practices, and industry standards requires constant attention. Governance frameworks need to evolve to ensure responsible, transparent, and ethical AI system development and deployment.

… People

People are at the heart of AI initiatives, from development teams to end-users. Team compositions change, expertise evolves, and new skills are acquired. Effective communication, collaboration, and stakeholder engagement are crucial. Additionally, user needs and expectations evolve, necessitating continuous research and design to meet requirements. Organizational changes and employee turnover impact continuity and knowledge transfer.

---

As you can see, there are easier initiatives one could imagine, but that doesn’t mean companies shouldn’t invest. The potential of AI for operational excellence and efficiency is significant. Companies that master AI will outperform their competition, with substantial impacts on their annual balance sheets.

⚠️ Given this complexity, we must be cautious in naming AI initiatives as simple projects. If approached incorrectly, the funding, timing, ambitions, and investment decisions might all be influenced adversely.

Common Understanding of AI Project

An AI project is a time-bound initiative focused on developing and deploying a specific AI solution to address a well-defined problem or opportunity. It involves clear objectives, a defined scope, and a set timeline. AI projects have distinct phases, including planning, data collection, model development, testing, deployment, and evaluation. The goal is to create a functional AI system within the allocated budget and timeframe.

So now, a straightforward question:

⁉️ How can we dare to name the majority of our AI initiatives simply an AI project, considering the highly volatile environment we are working in?

We are trying to build an AI solution that fits into business processes and employees' workflows, making them more efficient and effective. We need to guarantee that those AI systems remain stable given the volatile environment I have described above.

This means observing every dimension closely and continuously to react to or anticipate issues. But how can something be time- and budget-bound and still be continuous?!

Exactly, that's not possible!

In my experience as an employee and a freelance consultant, every single AI initiative has contradicted the definition of a project. Once deployed, new challenges inevitably arose, needing to be addressed. It often felt like a never-ending game of tackling these challenges, and this is a positive sign when you remain engaged. It indicates that your AI solution is worth maintaining, delivering enough value that reinvesting remains viable. However, there are scenarios where a single challenge can render the business case for your AI solution obsolete. At that point, it's time to manage the end-of-life cycle of your AI solution. For each challenge listed below, you must ensure that the investment to address the challenge keeps your business case profitable:

• Data:

  • Data sources frequently changed, necessitating continuous model updates.

  • New types of data were introduced, requiring adaptation of existing models.

  • Data quality varied over time, impacting model performance and requiring constant monitoring and cleaning.

  • Changes in data due to external factors made some AI solutions obsolete from a business perspective.

• IT:

  • IT infrastructure frequently evolved, requiring ongoing integration and compatibility adjustments.

  • Updates to security protocols or the addition of new data sources necessitated continuous modifications to the AI system.

  • Legacy systems needed to be integrated with new AI solutions, often causing delays and requiring additional resources.

  • Changes in hardware or software platforms required AI models to be re-engineered or optimized for new environments.

• AI:

  • AI algorithms needed regular tuning and improvement to maintain accuracy and performance.

  • New machine learning techniques became available, necessitating integration and testing.

  • Models required retraining as underlying data distributions shifted over time.

  • Continuous research and development were needed to stay ahead of the competition and incorporate the latest advancements.

• Business:

  • Business requirements and goals shifted due to market changes or internal strategic pivots.

  • New business processes or changes in existing ones required AI solutions to be adapted.

  • Shifts in customer behavior patterns necessitated retraining of models to ensure relevance.

  • Mergers, acquisitions, or changes in organizational structure impacted AI project priorities and scopes.

• Governance:

  • Regulatory changes and compliance requirements altered the scope and implementation of AI initiatives.

  • New data protection regulations restricted the types of data that could be used, requiring adjustments to AI processes.

  • Ethical considerations and governance policies evolved, necessitating continuous evaluation and modification of AI systems.

  • Internal governance frameworks required regular updates to align with best practices and industry standards.

• People:

  • Team composition and expertise evolved, impacting the development and maintenance of AI solutions.

  • New team members with different skills joined, or existing members received additional training, requiring adaptation in the approach to AI development.

  • Cross-functional collaboration needed to be maintained, ensuring that domain experts and AI practitioners worked effectively together.

  • Organizational changes and employee turnover required ongoing knowledge transfer and training.

By addressing each of these dynamic factors, you must continually assess whether the investment to address each challenge keeps your business case profitable.

Deciding Between a Project and a Product Approach in AI Initiatives

However, there might be scenarios where the above dimensions remain static, and approaching AI initiatives as simple projects might be appropriate. I haven't seen any, but here is a simple guideline to help you understand when a project makes sense and when a product approach is necessary:

Use a Project Approach When:

1. Static Environment: The data, IT infrastructure, and business requirements are stable and unlikely to change significantly over time.

2. Short-Term Goals: The initiative has a clear, short-term objective with a defined endpoint.

3. Limited Scope: The AI solution addresses a specific, well-defined problem with limited complexity.

4. Budget and Time Constraints: The project needs to be completed within a strict budget and timeframe.

5. Minimal Maintenance: The AI system requires minimal ongoing maintenance and updates post-deployment.

6. Proven Solutions: Existing, proven AI solutions can be easily adapted to fit the project requirements without extensive customization.

Use a Product Approach When:

1. Dynamic Environment: The data, IT infrastructure, or business requirements are likely to change over time, requiring ongoing adaptation.

2. Long-Term Vision: The initiative aims to provide long-term value and continuous improvement.

3. Broad Scope: The AI solution needs to address multiple, interrelated problems or evolve as new challenges emerge.

4. Flexible Budget and Timeline: The initiative allows for flexible budgeting and timelines to accommodate evolving needs and improvements.

5. Ongoing Maintenance: The AI system will require regular updates, monitoring, and maintenance to ensure optimal performance.

6. Customization and Innovation: The project involves significant customization or innovative solutions that need iterative development and testing.

Embracing the Product Mindset in AI Initiatives

But more importantly, for today's topic, let's revisit the question:

How can we dare to name AI initiatives simply an AI project?

It is simply not an AI project. Calling it a mere project is an insult to the efforts required to make AI initiatives impactful. We are not dealing with a single project; we are dealing with a lifecycle that requires numerous projects to maintain business objectives and meet user needs.

Perhaps this is why companies still complain about AI failures—their expectations are project-oriented. They believe that once a project is finished, success is guaranteed.

Well, bad news.

Simply naming an initiative a project doesn’t make it a project. 🤓

Adopting a product mindset means embracing the inherent risks and uncertainties that come with innovation. However, we all know that to unlock substantial rewards, you have to take risks.

You can minimize your risk by evaluating your initiatives across the six dimensions and performing thorough assessments of feasibility, viability, and desirability.

But I don’t want to force anyone to change their minds. It’s still your money, your targets, and your future, so go ahead:

Keep calling it a project. 😉

JBK 🕊

---

P.S. If you’ve found my posts valuable, consider supporting my work. While I’m not accepting payments, you can still help by sharing, liking, and commenting here or on my LinkedIn posts. This helps me reach more people on this journey, and your feedback is invaluable for improving the content. Thank you for being part of this community ❤️.

---