Forethought, a prominent player in the realm of customer service chatbots, is taking a leap forward into the realm of generative AI with the launch of "Autoflows." This innovative concept represents a significant evolution in the company's quest to streamline customer service processes.
Deon Nicholas, the co-founder and CEO of Forethought company, revealed the essence of Autoflows, describing them as "natural language simple prompts" designed to allow users to define their goals or policies, leaving the AI to handle the intricacies. Unlike conventional no-code or low-code workflows that necessitate explicit task definitions by customer service teams, Autoflows operate differently. Users can input a block of text specifying a policy, and Forethought's advanced AI, powered by SupportGPT, takes the reins. It draws from historical conversations and existing policy documents to discern the user's intent.
During its beta testing phase, Autoflows demonstrated an impressive boost in resolution rates compared to traditional workflows. Once equipped with the requisite data for a specific task, the AI autonomously engages customers, formulating coherent, human-like automated conversations.
SupportGPT, Forethought's generative AI model at the core of this system, is underpinned by a combination of OpenAI models and proprietary algorithms. While the process is largely automated, human oversight remains crucial for task addition, editing, or removal. The ultimate objective, however, is to empower the AI to handle as much as possible, aiming to reach a level of sophistication akin to ChatGPT's iterative refinement capabilities. Presently, Autoflows can integrate with systems like Zendesk to initiate tickets via a Rest API, although supplementary tasks necessitate manual input.
Forethought AI has recently opened the beta testing of Autoflows to the public, marking a pivotal moment in the platform's development. This move aligns with Forethought's mission to refine the system further and push the boundaries of customer service automation.
