Generative AI Framework: Orchestrate the Process

By Francisco Dagnino

Feb 01, 2024

Generative AI holds immense potential for organizations seeking growth, innovation, and a competitive edge. Achieving these goals, however, requires more than just selecting cutting-edge technologies. We need a structured approach and strategic implementation. That's where the "Orchestrate the Process" stage comes in.

This is the third installment of our six-blog post series on Generative AI strategy. In this post, we will take a comprehensive look at the "Orchestrate the Process" stage, which is crucial for success in implementing Generative AI strategies. We will explore how important it is to stay grounded in the overall business strategy. We will uncover the value of good data and key practices and steps that organizations, irrespective of their industry or size, can follow to drive success with Generative AI. Make sure to read the previous blog post Revolutionizing SMBs with Generative AI: A Comprehensive Guide Part 2

We'll also continue to discover how the imaginary CardboCanva, a leading cardboard packaging company, navigates the dynamic landscape of Generative AI implementation.

Orchestrating the Process: Instilling Generative AI as an Integral Part of Business Strategy

As with any major change, a good orchestration of the process can mean make or break for businesses. instilling Generative AI as an integral part of the business to support the overall business strategy is no different. Orchestration entails coordinating multiple moving parts without disrupting their normal functioning, thus ensuring a change factor integrates seamlessly with an existing operation to add value. Through well-orchestrated integration, Generative AI can become a transformative force, driving your business forward.

Read more on the topic on this blog post by our Chief Project Officer, Rodolfo Reygada, who has vast experience in leading complex industrial projects. He’ll be writing more on this and adjacent topics in the coming weeks.

1. Aligning to Business Strategy and Business Units

The effective orchestration of any new technology implementation must rely on careful alignment with the business’s strategy. Otherwise – I assure you – the cost and risk of expensive and irrelevant moonshots will quickly become prohibitive, hindering the adoption of technology that could truly provide a competitive advantage.

Regular checks against the business strategy enable businesses to ensure that the Generative AI initiatives are staying on track and are in line with the overarching goals and objectives outlined in the first pillar of this framework: Define the Opportunity. This alignment is crucial to avoid any deviation that could negatively impact the successful integration of valuable Generative AI initiatives into the business.

It's also important to ensure that everyone involved in the Generative AI projects has a clear understanding of the business strategy. Clear and regular communication of the business strategy among team members foster a shared understanding and interpretation, facilitating the integration of pilot and production projects.

Tools and Techniques to Stay Aligned to Business Strategy

All businesses will find their corporate culture to be quite unique, so there is no silver bullet to establishing a strong communication strategy. What I would recommend is to leverage multiple communication channels, rather than expecting a single email or post to do the job. Everyone responds to messages differently through different channels. Here are some ideas you can implement:

Workshops and Offsites: these are often fun and practical opportunities for different parts of the business to come together in an engaging way, priming them to perceive the vision behind a strategy.
Newsletters: with all their pros and cons, many find them useful and do read them.
Corporate videos and podcasts: put faces to the message. A CEO explaining their vision of a business is very different from an email.
Townhalls: there is nothing more palpable than human interaction, and a face-to-face stand-up can do wonders.
Success stories: leverage every little win and give credit to the relevant people. Make sure no one is left behind the curtains.

2. Risk Assessment

When instilling any change, conducting a rigorous risk assessment is paramount. No matter how promising the new paradigm can be, there are always risks associated. When developing a new Generative AI application, after validating that there is an opportunity for the business, the first and foremost aspect to address: data.

Data, data, data: Assessing the quality, trustworthiness, volume, sustainability, applicability of the available data is just the starting point, but it will help ensure one foundational input used for Generative AI applications is suitable and reliable. In a regulatory environment in development, understanding your foundations will allow the business to navigate legal changes certain to come. And soon. This will also ensure expectations of an initiative are properly managed, which will help retain attention from relevant stakeholders.
‍
Model Drift: in any AI or Machine Learning project, drift management is not only a risk, but a certainty – models’ behavior will evolve as the data input changes and the latter you cannot control. By closely monitoring data inputs, model performance, and outputs, businesses can proactively detect and address any discrepancies or anomalies.
‍
Third-Party Dependencies: running a generative AI model can get very expensive, very quickly. Computing and maintenance costs can make it prohibitive for most businesses to run on their own, so turning to commercial solutions like ChatGPT, Azure AI, AWS Machine Learning and a long and growing etc, is the obvious alternative. What if they unexpectedly go offline? What if they are deprecated? What if they are updated? Reliance on 3rd parties is a pillar of risk management, and thus should most certainly be included in any Generative AI initiative.

Though there is much more involved in risk assessment as a practice, these two items will cover much of the potential outcome of a Generative AI project. These aspects can affect the business’s strategy, image, customer experience, and employees’ performance. A holistic view of risks allows businesses to navigate the challenges of change and ensure the successful integration of Generative AI while safeguarding their overall business interests.

3. Developing a Use Case Framework

The value of a well-defined Use Case cannot be underestimated. In the process of assessing whether to invest in a particular initiative – no matter the size or context – proper assessment of the expected impact is often overlooked. Enter Use Case Frameworks.

A use case framework allows parties involved to stay aligned with aspects of project outcome that a business considers important. A use case framework as part of a Generative AI Strategy outlines the boundaries of applications of the technology within the business’ operations, addressing key challenges or leveraging opportunities to achieve desired outcomes. It aligns with the business strategy, Generative AI strategy objectives, risk assessment, and more.

By developing a bespoke use case framework, businesses ensure that their Generative AI initiatives are aligned with their business strategy and objectives. It gives them a structured approach to identify and prioritize use cases that provide the greatest value. The framework also helps organizations evaluate the feasibility, cost-effectiveness, and potential risks of implementing each use case. By carefully selecting and implementing use cases, businesses can fully integrate Generative AI into their operations, maximizing its impact on business growth and innovation.

Pointers to define a Use Case Framework

Use cases are a powerful tool to communicate expectations as they ultimately must address the questions: what challenge (framing a problem as a challenge has multiple advantages) am I solving? Why is this important? For whom am I solving it? How will I solve it? A well-defined use case can have multiple dimensions and it is up to the business to identify what is relevant provided a particular context. Here’s a list of items you can include in a Use Case framework:

Name it: a single phrase that describes the challenge and approach. A common structure is “Leveraging [A TECHNIQUE] to address [A CHALLENGE]”
Challenge statement: succinctly define the problem to be addressed.
Impact of solving the challenge: why is this important?
Stakeholders: list of impacted stakeholders and how they are impacted by the challenge
Solution outline: brief of the solution – don’t get into technical details
Solution detail: at-length description of the solution’s architecture, avoid any specific technology if possible and stick to broad areas (e.g. generative AI, real-time visualization, edge computing)
Identified Risks: a comprehensive list of identified risks.
Potential Risks: a list of areas that could be impacted. Outline potential scenarios, even if fringe.
Resources: a list of needed resources, be it infrastructure, 3rd party vendors or consultants, licenses, etc.
Sponsorship: list down the leadership entities required to sponsor the solution for success.

CardboCanva Case Study: Orchestrating Generative AI for Growth and Innovation

In their journey of orchestrating Generative AI, CardboCanva started by identifying key opportunities to leverage this technology. Building on their previous successes, they pinpointed three specific areas where Generative AI could drive growth and innovation:

Customer Packaging Design Assistance
Automated Specialized Customer Support for Top-Tier Customers
Carbon Footprint Analyzer for Customers

Learn more on my previous post here.

Aligning Business Strategy and Business Units

To effectively and continuously communicate their Business Strategy within all stakeholders, CardboCanva structured a number of spaces all with different cadences and stakeholders:

Monthly Business Strategy Updates: regular reviews and updates led by business unit leaders to report on evolving market demands and changes in customer preferences. This served as a space for the CEO to regularly remind leaders of the key aspects of their business strategy.
‍
Strategic Alignment Sessions: quarterly reviews to understand the scope of different Generative AI-led ideas proposed and question their alignment with the overall business strategy.
‍
Performance Monitoring and Evaluation: CarboCanva implemented a scorecard that helps level the measured impact of any Generative AI initiative instilled, using a combination of objective and subjective metrics captured from IT, the GenAI team, user feedback, and employee feedback.

Risk Management

CardboCanva decided the best way to maintain a sustainable approach to empowering innovation without the shadow of risk management was to minimize the uncertainty of proposing a new idea by rewarding novelty beyond safety.

Risk Assessment Framework: CarboCanva developed a comprehensive risk assessment framework specifically tailored to Generative AI projects. They identified potential risks at different stages of and provided clear guidance on how to manage such risks, thus reducing the apprehension of proposing novel ideas.
‍
Risk Mitigation Strategies: To manage and mitigate risks, CarboCanva implemented semi-automated and Generative AI-led processes to assess data quality, algorithm biases, regulatory compliance, and potential negative impacts on customers, users, and employees. Adopted overarching ethical AI principles, adhered to data protection regulations, and regularly conducted audits and evaluations to ensure compliance.

Continuous Monitoring and Adaptation: CarboCanva implemented continuous monitoring processes to detect any deviations or emerging risks, leveraging multiple tools and automations. They regularly reviewed and updated their risk management tactics to address new challenges and changing circumstances.

Structuring a Bespoke Use Case Framework

To guide their Generative AI initiatives, CardboCanva decided on defining principles to a use case, rather than a standard format. This was in response to the strong resistance they experienced in early stages of communicating the new strategy:

Alignment with Business Strategy: CarboCanva ensured that every use case identified and pursued aligned with their overarching business strategy by catering idea capture to the different natures of talent and background of their business units. Ultimately, assessing idea potential to contribute to key business objectives, drive growth, enhance customer satisfaction, or support sustainability goals, was left to a dedicated multi-disciplinary council.
‍
Resource Allocation: with limited budget, CardboCanva decided on prioritizing speed for "low hanging fruit" use cases, while thoroughly documenting use cases with high potential. The latter were brought to a specialized consulting firm for further validation. This workflow ensured all ideas were placed in the right bucket: execute or study further.

Conclusion

The beauty of the word Orchestration lies in the concepts of integration and interpretation, and the obvious metaphor with music. Just as an orchestra Maestro, a good orchestrator will leverage the plan (sheet music) and the moving parts of a business (musicians), with all their nuances, roles, functions and constraints (the instruments) to integrate their potential and ever so slightly modify their behavior to achieve its interpretation of the opus: the Objectives of leveraging Generative AI at the service of the business.

Orchestrating the Process of instilling Generative AI into the fabric of your business will enable you and everyone involved to maintain perspective for the value of the technology supporting the business strategy.

‍