Adoption of Generative AI
ChatGPT and generative AI have taken the world by storm. By far, it has been the fastest adoption of a new technology. What does this mean for an enterprise, and how is this shaping the product roadmap we are building at Yellow.ai?
Massive consumer adoption is a leading indicator for enterprise adoption:
- Google search -> Enterprise search
- Consumer mobile apps -> Enterprise mobile apps
- Consumer messaging (WhatsApp, Messenger) -> Enterprise messaging (Slack, Teams)
Fully expect conversational interfaces for enterprises to further accelerate and skyrocket in mainstream.
Large Language Models
LLMs fine-tuned on proprietary enterprise and domain data will gain enterprise adoption. Language models like BERT and GPT-2 have already been leveraged in advanced conversational AI systems like Yellow.ai primarily at understanding conversations and generating training content for new intents. However, with the advent of extremely large language models (175B+ parameters), these systems have demonstrated the ability to generate verbose human-like text. The incremental cost of generating new content rapidly comes down.
Enterprise use cases
In an enterprise context this will drive use-cases such as generating support articles, hyper-individualized marketing campaigns (imagine unique content for each user), HR policies. The general-purpose LLM will not work for these but will need domain-specific LLMs that are built or fine-tuned on massive amounts of proprietary data. Yellow.ai is deeply investing here to enable dynamic content creation for a comprehensive experience with enterprise chatbots and virtual assistants.
Enterprise conversational use-cases go beyond just text generation; they need to train intents as well as execute actions and workflows as part of conversations. One of the most complicated tasks in creating a NLU bot is training the intents. With our proprietary DynamicNLP, we have already lowered the barrier for training common intents. By using generative AI, we will lower the barrier even further and enable anyone to train custom intents that are specific to their business. Further, we will see generative AI build, stitch workflows and create required actions on demand. Currently, workflow creation and integration in conversational systems is manual and rigid, even if they are created with integrated no-code tools. This is an area of generative AI that is still unexplored, and the Yellow.ai roadmap is working toward enabling dynamic creation of workflows embedded within conversations.
Autonomously improving conversion optimization using multi-variant text optimization at scale
The extreme low cost of generating versions of interactive text will lead to companies dynamically generating multiple variants of text for conveying the same information or taking the same action and use reinforcement learning to optimize for the variants that lead to the best conversion. We will see support and marketing interactions autonomously and continually improve conversions with continuously improving text variants. Clearly, this is on our roadmap.
No more stale, similar-sounding robotic voices when you call customer support. Be prepared for dynamically changing personalized voices coming to your support that will take the frustration out of the support experience and make your consumers love support. Generative AI will help deliver this.
We can’t be more excited about the pace at which AI is evolving and the impact it will have on how consumers and employees interact with businesses.