The Potential of Generative AI for Accelerated Business Transformation

How is GenAI different from AI?‍

Traditional AI processes data to make predictions, while GenAI uses existing data to produce synthesized content.

‍Why is GenAI so popular?

‍While AI has been in our lives for over a decade, most of its applications require explanations to be appreciated. That is not the case for GenAI, which produces immediate results, easily processed, even easier to enjoy – but also more daunting.

GenAI’s purpose is not to imitate a human brain and thus replace human involvement or meant to do everything we do. Some AI applications go beyond the human scope, such as converting a 2D image to 3D; others are performed better, like identifying certain cancerous cells in an MRI or playing chess. However, in most cases, GenAI performs unremarkable tasks in remarkable ways: adequately or about as well as a trained person would. Such is the cast for translation, code writing, speech-to-text, and the like, which are simply rendered more accessible. But with the right expertise and some creativity, a few less-discussed applications of GenAI could also accelerate business transformation in new, unexpected ways.

At 55, we’re no stranger to GenAI, applying this technology to data analytics and change management, amongst many other examples. Below, we’ll explore six use cases illustrating how 55 and Brandtech partners create innovative solutions to common digital challenges, and how to deploy these technologies within your organization with details gleaned from our recent webinar, The Potential of Generative AI for Accelerated Business Transformation. This webinar featured four of our French office’s AI experts: Pierre Harand, Partner & Co-Ceo; Jean-François Wassong, Partner & CTIO; Tiyab Konlambigue, Cloud Expertise Director, and Bastien Chappis, Data Scientist Lead.

What can GenAI be used for?

DATA ANALYTICS

‍Data is central to agile decision-making within any company. For better KPI visibility, queries, dashboards, and graphics creations are par for the course, but such processes can require prolonged back-and-forth between teams. Here, 55 applies GenAI to translate natural language into SQL queries and graphs within seconds.

‍DATA INTELLIGENCE

‍Between our current “big data” era and increased cloud capabilities, vast amounts of data can be stored. Through our Generative AI solution, you can now sort out useful data from the useless, as 55’s tool identifies patterns for you.

‍CHANGE MANAGEMENT

‍Using GenAI to facilitate the adoption of new tools within your organization, this solution provides on-demand support, training, and resource access. The Libsearch engine extracts and summarizes findings from internal and external databases to offer a fully interactive library, FAQ, and “ask me anything” function.

‍CAMPAIGN TAXONOMY

‍To ensure accurate marketing campaign reporting, which tends to feature errors and could thus lead to incorrect decisions, this 55 tool finds mistakes in taxonomy, proposes corrections, and allows constant monitoring for optimal media strategies.

‍PRODUCT FEED IMPROVEMENT

‍Following a similar process, 55 deploys Generative AI to product feed files to correct, complete and unify their content; this application allows every item to be correctly referenced and thus ensures its availability to potential customers on appropriate channels, improving digital sales.

‍CONTENT PRODUCTION

‍Part of the Brandtech group, Pencil can create short video ads for web marketing campaigns in just a few clicks. After uploading your logos and graphic resources, write a brief like you would for a creative agency (message, audience targeted, ...), then select a specific channel, i.e., Tiktok or Instagram. Pencil will generate a tailored web campaign based on your input; you can then edit it as needed for a perfect fit.

How to deploy Generative AI

Four essential steps for a safe GenAI deployment

The 55 methodology

As Generative AI is still relatively new, lack of access to the right technologies and talent is currently the most significant hurdle faced by businesses willing to adopt GenAI solutions. Armed with experience and a team of experts, 55 has built a methodology to implement such solutions seamlessly and successfully.

First, through a deep understanding of the technology itself, layer by layer, 55 applies a “top-down” approach to GenAI implementation.

Each solution must answer a business need. Our varied expertise is applied to determine the correct solution for each requirement; technical applications are then based on custom models or default foundation models from CloudAI, Google, etc. These models rely on both data and MLOps strategies, and the overall architecture sees its services provided by partnering cloud platforms.

Once the technical architecture is established, the framework phase of the project can start. This phase is comprised of three major steps:

Discovery Use Case Definition & Ideation

‍Business opportunities are identified, use cases are defined, and pilot implementation path is determined through 55-led workshops.

‍This phase is centered around considering needs and figuring out which areas would benefit the most from GenAI implementation: typically, process automation, insight generation, and user engagement. Once potential bottlenecks are identified, data availability and internal skills are assessed, and a clearer picture of scaling challenges emerges (which teams would be affected and when amongst other considerations).

Once opportunities are identified, 55 can determine use cases, keeping cost/benefits, technical and organizational constraints and potential impacts in mind.

Typical use cases:

• conversation (intranet Q&A, customer service, website navigation…),
• content generation (creative assistant, document generation, developer assistant,...)
• data analysis (ad-hoc data analysis, product & content recommendation, compliance & regulations…)

PoC (Proof of Concept) & Proof of Value

Define the boundaries of your model and how it’s fed and monitored. After developing a pilot solution, demos are provided to stakeholders to increase enthusiasm, and a roadmap to production is created.

‍To determine what constitutes the core of your model, you need to select a base (translation, classification, summarization, etc.) and determine the expected output of a prompt. Optionally, you can also decide on the size and volume of said model, as it will affect cost, and fine-tune it as necessary.Incidentally, data may need to be prepped, and questions of safety and reliability must be addressed; data security is essential for a successful implementation, as all teams might not need the same level of access, as is rooting out biases early on. Monitoring is set up through user feedback, and deployment is finalized.

Scale Up

Make your model production ready. Launch your solution, and monitor the success metrics determined during the discovery phase. You can then potentially identify your next use case and start the process again as needed.

‍As GenAI is still in its infancy in many respects, scaling up can prove quite challenging. Taking all considerations into account is critical, considering all of your possible constraints and limitations, from inconsistency to ambiguity, security, and cost. No need to panic – proper integration is, indeed, possible! But to get there, a high level of collaboration is required between technology experts and project owners.If you want to know more about how fifty-five can help you adopt and master the GenAI application that best matches your needs, please fill in our contact form for more details.