Generative AI applications and their underlying models, such as GPT-4 and ChatGPT, have deservedly received much attention lately. They are some of the latest in the wave of innovation using AI and large language models (LLMs) in recent years. As the models have become bigger and bigger – into tens and hundreds of billions in parameter size – and the data feeding into them has grown massively, they are getting better at a variety of tasks, like information extraction, summarization, multi-turn conversations, text/image generations and lately, even reasoning and instruction-following capabilities.
The foundations of this technology are well established. SymphonyAI has used large language models since our inception, starting with the early transformer-based encoder-only models like BERT and RoBERTa for natural language understanding (NLU) and the early transformer-based decoder models like GPT2 for natural language generation (NLG) in numerous AI-driven applications. In the last year, SymphonyAI developers have fine-tuned open-source variants of GPT, like GPT-J and GPT-NeoX, to create domain-specific models that serve as foundational models for our divisions’ applications.
SymphonyAI Retail CPG and SymphonyAI Summit’s digital assistants use these technologies most strikingly. Serving as copilots for frontline workers, managers, and executives, these enterprise AI solutions are at the forefront of this new era where business leaders are “taking assistive technology to a new level, reducing application development time, and bringing powerful capabilities to nontechnical users,” as McKinsey wrote in a recent report.
SymphonyAI Retail CPG’s CINDE digital assistant generates new sales and greater shopper engagement, recommends critical actions, and frees up time for strategic decision-making in retail stores. Without such a tool, brand managers today must usually request specific sales information from analysts, a cumbersome process. With CINDE, they have direct access to real-time sales figures, store data, and other reports through natural language queries.
SymphonyAI Summit’s digital assistant for IT service management provides users contextual answers to frequently asked questions, like employee inquiries on benefits or time-off policies, fixes for routine IT problems, and even equipment and digital asset tracking. This digital assistance streamlines users’ work and frees up the IT team to focus on more involved tasks.
Both digital assistants improve business operations and processes while generating and organizing data for business insights. Each additionally offers transparency and explainability to track the inputs and decisions behind AI’s performance to provide invaluable information on optimizing businesses.
AI-driven SymphonyAI applications illustrate how predictive AI models can drive immense business value and co-exist with generative AI models to improve how users interact with these technologies in their domains.
Just the beginning
SymphonyAI teams are excited about the rapid pace of innovation in generative AI, including new foundational models with better instruction-following capabilities, better reasoning and explainability, and broader applications for many different tasks across industries. These approaches will be part of the new generative AI products our verticals will unveil later this year.
ChatGPT is just the start. The next phase of AI applications will be even more exciting. Business leaders should be looking at AI now – or risk falling behind during this remarkable period of innovation.