Editor’s note: Jason Tan is a speaker for ODSC APAC this August 22-23. Be sure to check out his talk, “Framework and Lessons Learned from Building a Generative AI Application,” there!
In the dynamic realm of AI, the introduction of ChatGPT on November 30th, 2022, marked a monumental turning point, reshaping human-technology interactions and shaping AI’s future. Tech giants like Google, Facebook, and Amazon raced to develop their versions of ChatGPT, indicating the revolutionary impact this innovation had on the AI landscape. This paradigm shift extends beyond industry giants, captivating startups and small tech companies, leading to Generative AI integration into their software and reshaping customer experiences and operational models.
As the significance of ChatGPT extends, it prompts discussions about its implications across diverse industries, including the corporate world, higher education, and government agencies. This interest has even caught the attention of business stakeholders, engaging their Chief Data and Analytics Officers (CDAOs) to explore Generative AI’s use cases and its potential with Large Language Models (LLMs) for their enterprises.
In this article, I present the first step in a proven framework, born from my own experiences and successes with Generative AI, vector databases, embeddings, and LLMs. This framework encompasses the development of a Conversation CoPilot, assisting technical founders in establishing relationships and driving efficient sales, and the deployment of Generative AI for enterprises.
Step 1: Identify Business Use Cases
One major pitfall organizations must avoid is the Shiny Object Syndrome (SOS) — making tech adoptions without a clear understanding of their implications. CDAOs must collaboratively explore the potential use cases, ensuring alignment with core objectives, priorities, and long-term goals. Use cases might involve improving existing customer service chatbots, enhancing knowledge management, refining marketing and sales copy, or deploying Conversation CoPilots.
Step 2: Decide the Minimum Viable Product (MVP)
Collaborating with experts well-versed in Generative AI and LLMs, organizations can craft an MVP to address specific business goals. For instance, an organization might deploy a vector database to store unstructured data, employ embedding techniques to query data, and integrate LLMs to contextualize company policies for customer replies. This approach streamlines operations and enhances customer interactions.
People remain pivotal in even the most advanced AI systems, as quality data is integral to success. Therefore, it’s crucial to engage the right stakeholders, both internally and externally, throughout the process. This ensures a well-rounded approach and bridges the gap between analytics-driven insights and actionable outcomes. Simultaneously, organizations must identify the required data and systems for successful implementation.
Step 4: Develop Implementation Strategies
The selection of an appropriate generative AI application strategy is essential for deployment. Organizations must decide whether to leverage an existing software development kit (SDK), partner with external consultants, or build an in-house solution. Each option has its merits, depending on the organization’s capabilities, resources, and strategic objectives.
Step 5: Transform Insights into Action
Finally, organizations must focus on translating the insights generated by Generative AI into actionable outcomes. This involves integrating the AI outputs into existing workflows, refining decision-making processes, and improving operational efficiency. The seamless integration of AI recommendations into established systems empowers employees to make informed decisions and fosters an environment of continual improvement.
As the AI landscape continues to evolve, embedding Generative AI into enterprises emerges as a game-changer. By adhering to this first step as part of a comprehensive framework, businesses can make informed decisions, avoid the pitfalls of Shiny Object Syndrome, and harness the true potential of AI to transform their operations and customer experiences.
To dive deeper into the intricacies of this framework and learn more about the next 4 steps, I invite you to join my upcoming talk on this topic at ODSC APAC 2023 on the 23rd of August. I will also share invaluable insights gained from real-world experiences, challenges faced during development, and the lessons learned from overcoming obstacles. This is your chance to explore the transformative power of Generative AI firsthand. See you there.
Jason Tan is the Founder of Engage AI, a Conversation Copilot that remembers conversations across multiple channels to augment conversations in virtual and real-life. Since its release in Jan 2023, over 30,000 users worldwide have been using it to break the ice and engage with their prospects. Taking the learnings from implementing Engage AI, he also assists and shares the learned lessons with enterprises to embrace and incorporate Generative AI and Large Language Models into their business.