Generative AI is not a new concept but since it has entered the public domain, evolving into something capable of reshaping the digital landscape, businesses all around the world are trying to tap into unprecedented opportunities for innovation and efficiency.

At Metyis, we've been investigating this technology for many years and with the expertise of people like Louis de Cointet, a Partner based in our Paris office we are guiding our clients through the complexities of GenAI adoption and implementation.

The era of GenAI expertise at Metyis

With over 20 years of experience in designing innovative digital solutions, Louis de Cointet brings a unique blend of technical knowledge and perspective to our GenAI initiatives. 

Louis reflects on his journey: "I've been deeply immersed in this field since my academic days in Technology and Finance. The startup I founded in 2017, which became part of Metyis in October 2022, specialised in chatbots and had a strong focus on AI." 

Louis and his team's early adoption of GenAI technologies has given Metyis a competitive edge. As Louis notes, "Our journey into Generative AI began naturally in 2019-2020, even before the term was widely recognised, as we explored models and architectures to enhance our solutions. We even tested GPT-2 in 2021 for a client's project, which provided us with early insights into the potential of these technologies."

The potential of GenAI across business functions 

The versatility of GenAI allows for its application across various business functions. Louis highlights the rapid deployment capability of GenAI: "When this technology is tailored to the specific needs of a customer, it can be deployed far more rapidly than traditional AI approaches, especially in areas like complex text analysis, message classification, and, of course, content generation." 

He adds, "This enables us to implement conversational solutions, such as chatbots, with remarkable speed, delivering a level of response quality that far surpasses what was possible with earlier tools."

Metyis' integrated GenAI approach 

As with all our digital and data solutions, Metyis’ approach to GenAI implementation is comprehensive and client-centric, aiming to sustainably integrate all aspects of our partner’s existing processes.

Louis explains our process which has become best practice across all teams: "Always begin with the use case—whether it's content creation assistance, customer chatbots, or text corpus analysis, to name a few. Once the functional requirements are clearly defined and validated, identify the key constraints, such as data, architecture, security, intellectual property, user interface, and API integration. Be clear about the KPIs you want to achieve and how you will measure them. This approach allows for the creation of a detailed specification that outlines both the technical architecture and the AI pipeline, paving the way for effective development." 

He emphasises the iterative nature of the process: "And, of course, you can expect to iterate several times on the result to find the right solution, thanks to testing and feedback from users." 

When it comes to the technical aspects of developing custom GenAI solutions, Louis stresses the importance of a deep understanding of the business context: "Truly understanding the business need and the surrounding environment is crucial, as these factors will dictate the choice of model(s) and the pipeline to be implemented."

He further elaborates on the rapidly evolving landscape of Large Language Models: "This is especially true with LLMs, where the rapid evolution of models on the market means that cost and performance can vary significantly depending on the specific use case. Selecting the right model is essential to achieving optimal results. I have in mind a project where we observed a 1 to 10 cost ratio between two models designed to address the same need."

Selecting the right LLM by creating a tailored solution is something Louis considers to be of the upmost importance and value, "The starting point must always be the functional use case, the business objectives (including ROI), and the client's context, such as data and technical environment. Large Language Models (LLMs) can be an ideal fit for certain needs, but they may not always be the best solution." 

He cautions against over-reliance on off-the-shelf solutions: "It's crucial to maintain a critical perspective, especially as it's becoming increasingly easy to believe that a solution can be delivered 'turnkey' through an API, which is rarely the case." 

Balancing innovation and reliability

While the potential of GenAI is immense, it will always be important to have an awareness of the need for responsible AI development. Louis emphasises this point: "First and foremost, it's important to remember that while LLMs have revolutionary aspects, they come with significant limitations. Notably, they lack true understanding and reasoning capabilities, which can result in what is known as 'hallucination.'" 

He adds, "Moreover, the costs to run these models can quickly become difficult to bear if you are processing or generating large amounts of complex data, not to mention image and video." 

As a way of tackling these challenges, we've developed robust methodologies for performance measurement and oversight. As Louis explains, “To address this, it's essential to design systems where performance measurement—whether through statistical tools or human oversight—is central to the approach.”

The future of practical GenAI implementation

When asked about successful GenAI implementations, Louis provides insight into his current work: "Rather than focusing on a single project, I'm working on several initiatives, all aimed at delivering high-quality information through conversational services to both internal and external users, leveraging company data." 

He highlights the importance of continuous improvement: "As these solutions evolved, we had to release multiple versions to accommodate a growing number of testers and to continuously refine the data that powers responses and enhances the user experience. It's crucial to remember that these tools are strong indicators of your data's quality!”" 

Looking ahead, Louis sees a shift in the skills required for working with GenAI: "In all cases, it's essential to have a deep conceptual understanding of the approaches and models involved. As we increasingly rely on complex pre-trained models, such as LLMs, the focus shifts more toward prompting and fine-tuning these 'black-box' models, often at the expense of traditional machine learning training methods." 

He adds, "Data scientists of the new generation are likely more 'polyformal' than ever, possessing strong skills in cloud computing and data engineering, in addition to their traditional background. This is because all these capabilities are now more interconnected than ever before." 

Expertise leads the way

In the new era of GenAI, Metyis stands ready to guide businesses through every step of their journey. Our expertise, strategic approach, and commitment to responsible AI development position us as a trusted partner in unlocking the transformative potential of GenAI. 

Louis offers this advice to organisations looking to adopt GenAI solutions: "AI, especially Generative AI, is not magic, but it can deliver substantial benefits to many of your needs, and it can do so rapidly compared to traditional approaches. However, to fully harness its potential, it's crucial to have the support of expert 'doers' who have successfully implemented these solutions before—something we, Metyis, bring to the table." 

The path to GenAI success is complex, but it is well within reach with the right guidance, expertise, and strategic approach.