Impactful AI Solutions: A Five-Phase Framework for Project Scoping

Create a tangible and meaningful impact with our five-phase framework to seamlessly align AI project scoping with your business objectives.

Venturing into the intricate realm of AI project planning often feels like navigating a labyrinth of possibilities, just as complex as the algorithms themselves. The allure of the latest machine-learning techniques is undeniable, but without a well-structured approach, you risk getting lost in the technological maze. In the world of AI, success hinges not only on technological prowess but also on aligning your solutions with the broader business strategy.

Whether you are a product owner defining a business problem, a business analyst evaluating proposed solutions, or a project manager tasked with allocating resources, having a defined, business-driven scope and plan is crucial.

This comprehensive, five-phase framework seamlessly aligns AI project planning with your business objectives to create a tangible and meaningful impact.

Phase One: Identifying the (Business) Problem

In today’s tech-savvy landscape, it’s all too easy to get swept away by the promise of cutting-edge AI solutions. But before diving in headfirst, take a step back and ask yourself: What precisely are we solving? Failing to address this fundamental question often results in solutions searching for problems, leading to squandered resources and unmet business objectives.

Take a healthcare project, for instance, where a potential problem could be high readmission rates. In such a scenario, we can tailor AI to predict readmissions, offering a targeted solution that enhances patient care while effectively cutting costs and improving the patient’s well-being.

The next logical step involves identifying key stakeholders. In our example, obvious stakeholders include healthcare providers, patients, and insurers. However, the nuance lies in recognizing hidden stakeholders. These are parties indirectly affected by the project, such as local communities, adjacent industries, regulatory bodies, or, in the example of healthcare, even medical researchers.

We could only observe the effects on some of these groups in the long term. Ignoring these hidden stakeholders could result in unexpected roadblocks, ethical dilemmas, or missed opportunities for broader impact. A thorough stakeholder analysis enriches the AI project by making it more comprehensive, ethical, and aligned with the overarching business needs and societal values.

Questions to Answer:

1. What precisely are we solving?
2. Who are the key stakeholders?
3. Who are the hidden stakeholders we haven’t considered?

Phase Two: Exploring Diverse Solutions

With a clearly defined business problem, the following strategic phase explores various potential solutions. This is especially crucial for roles like Product Owners and Managers, who must look beyond the first solution that comes to mind.

It’s tempting to latch onto the first seemingly viable idea, but this narrow focus limits the potential of your AI project. For example, if the problem is predicting patient readmissions in healthcare, one approach is to analyze electronic health records, while another might involve real-time monitoring data. Furthermore, it’s essential to compare the benefits of using a pre-trained model, if applicable, or training one from scratch. Time and budget constraints play a crucial role in this phase, affecting the selection of alternatives.

Diving deeper into ideation also prompts you to question the robustness of your data. Would your solution benefit from combining electronic health records with real-time monitoring data? More diverse and voluminous data sets can improve model accuracy but may require additional processing power and expertise.

Another avenue for exploration is transfer learning, a technique where a pre-trained model is adapted for a new but similar problem, akin to the recent trend of using large language models (LLMs). Transfer learning is both a time-saver and a resource optimizer, allowing you to leverage pre-existing solutions for your business case. On the other hand, a customized model offers a more tailored solution, addressing niche nuances.

Questions to Answer:

1. Do we need a pre-existing model or a custom one?
2. Could more data enhance performance?

Phase Three: Assessing Feasibility and Added Value

After brainstorming an array of AI-driven solutions, Business Analysts and Data Scientists step into the spotlight. It’s time to evaluate the feasibility and added value. Remember, it’s not just about building a proficient AI model, it’s about contributing tangible benefits to the problem you’re tackling.

In our example, running a few pilot tests using historical patient data is instrumental in validating if your AI algorithms meaningfully improve readmission predictions. Whether it’s increased predictive accuracy or a quantifiable reduction in readmissions, the metrics should resonate with the key concerns of your identified stakeholders, ensuring the project remains aligned with its core objectives.

Feasibility and value aren’t the only metrics for success — ethical dimensions must also be front and center. Beyond ensuring data privacy and obtaining informed consent, it’s crucial to take measures to prevent your AI model from perpetuating existing societal biases. This means looking after people’s data and ensuring the AI doesn’t unfairly favor any group. So, in addition to asking if your idea is possible and helpful, it’s pivotal to ask if it’s ethical and fair.

Finally, it is critical to understand and plan for compliance with data regulations such as GDPR, especially for global operations.

Questions to Answer:

1. How do we measure success?
2. What ethical safeguards must be instituted?
3. What regulations do we need to comply with?

Phase Four: Setting Clear Objectives

The next phase defines what success looks like — the domain in which Project Managers and Product Owners will feel at home. For healthcare, it could be reducing patient readmissions by 20% while maintaining accurate predictions. These concrete goals keep everyone aligned and allow for progress measurement. But don’t just define what you want to achieve. You must also map out how you’ll get there.

Consider setting deadlines for data collection, model training, and evaluation. Adopting an agile approach helps your team remain flexible and responsive, which is especially crucial in AI projects where initial assumptions change quickly. Understanding what you’re targeting and how you plan to get there ensures a well-rounded and effective AI project.

Questions to Answer:

1. What is our execution timeline?
2. How do we plan to collaborate effectively?

Phase Five: Allocating Resources Effectively

With your AI project’s goals and plans in place, you need to allocate the necessary resources before proceeding. This goes beyond data and algorithms. In our healthcare example, a multidisciplinary team might be necessary, encompassing data scientists and medical professionals for domain expertise and bioinformaticians for data engineering. It’s essential to be aware and clear on whether you have the right mix of expertise, data, and perhaps even specialized hardware like GPUs for complex model training or dedicated servers for real-time data processing.

However, resources are more than a checklist. They’re an evolving set of needs and costs that can change as your project progresses. It’s vital to anticipate both the upfront costs, like model training, and ongoing expenses, like data storage or additional software. Your budget should not be set in stone as it’s a living part of your project that likely needs adjustments along the way. A deep understanding of your resource needs and how they translate into costs often separates a successful AI project from one that struggles to deliver on its promise.

Questions to Answer:

1. What types of resources are essential for this project?
2. How will the resource needs change over the life of the project?
3. What is the estimated budget, including upfront and ongoing costs?

Conclusion

Successfully planning and executing an AI project is an ongoing, iterative cycle that demands a perfect alignment between cutting-edge technology and business objectives. This five-phase framework is more than a set of guidelines — it’s a robust strategy designed to be your cornerstone in this complex endeavor. Adaptability is vital, so prepare to refine your approach based on fresh insights and constructive feedback as your project evolves.

Armed with this framework, you’re not simply planning an AI project but paving the way for transformative business solutions. The next step is yours to take, and it could turn your AI vision into a reality. Looking for ways to speed up the AI development process? Stay tuned for our next article to explore the game-changing concept of prompt-based development.

Original post here.

Authored by Georgi Naydenov and Bobby Bahov.

About Georgi Naydenov:

Georgi is a versatile professional with deep expertise and a broad understanding across multiple areas. His proficiency lies in his capability to identify, analyze in depth, and synthesize data to allow for the maximization of business value. His proficiencies extend to both healthcare and energy analytics.

His enthusiasm for data and AI is unwavering, particularly in the prompt engineering area. Over the past year, he’s been actively researching ways to integrate foundation models, such as GPT-4, Vicuna, Claude, Falcon, and Llama 2, into his daily workflows and projects. Georgi is enthusiastic about providing guidance on mitigating language models’ hallucinations and optimizing their efficiency, thereby enabling the implementation of the models’ output for tangible product development. In his leisure time, Georgi is infatuated with metalcore concerts and mosh pits that complement his journey of self-learning the guitar.

About Bobby Bahov:

With more than a decade of experience in software product development, Bobby seamlessly navigates the intersection of cutting-edge technology and business efficacy. Starting his career as a software developer, he quickly advanced into project and product management roles. He has managed projects and consulted a varied clientele — from startups to large corporations, as well as governmental organizations and NGOs. He has also spearheaded multiple innovative ventures and programs focused on AI tools, satellite data, and space technology.

An alumnus of the Rotterdam School of Management with a Master’s degree in Business Information Management, he’s currently pursuing a Ph.D. researching AI simulations, digital twins, and synthetic data. Today, as a Senior Business Analyst at Mentormate, he continues to be a driving force for innovation — and make no mistake, he’s still one of the biggest geeks you’ll find around.