No ratings yet.

Generative AI offers many opportunities for organizations to make their customer support processes more efficient. While the digital assistant can answer common questions and execute routine tasks, support agents can focus on more complex queries. However, this also brings strategic challenges for companies. To what extent should companies automate customer service processes, and how can they prevent losing the personal touch in these processes?

Introducing HILAI

Our team focused on these strategic issues by proposing HILAI (Human in the Loop AI), an autocompletion feature for customer support agents to combine the advantages of genAI with human interaction for Tidio. Inspired by AI coding copilots like Cursor and Github Copilot, HILAI acts as a customer service co-pilot. It leverages advanced LLMs and NLP to analyze conversation context and proactively suggest responses to customer service agents. Agents can quickly implement these suggestions, maintaining human oversight throughout the interaction.

HILAI’s development is projected to launch in 2025, with a 6-12 month rollout plan. The project is divided into three phases: Development, Testing & Feedback, and Training & Deployment. Key performance indicators will include AHT reduction and CSAT improvement.

Figure 1
Business Model Tidio as-is

Value proposition

Figure 2 describes one of the main challenges companies face when implementing AI tools within their customer service processes. HILAI takes these decisions away from the strategic level to the customer agent who can decide to what extent empathy and customization are needed versus the advantages of AI, based on the context of each conversation. An interviewed company mentioned advantages such as faster response times, more consistent on-brand messaging, and the ability to handle a higher volume of tickets efficiently. However, disadvantages may be agents relying too heavily on AI assistants and the needed human review for generated messages. Summarized, the value proposition of HILAI consists of cost-effective AI-powered customer service with human oversight, enhanced agent productivity and happiness, and reduced ticket handling time.

Figure 2
Digital CX Pyramid

Note. From “AI-Powered Customer Service: Does it Optimize Customer Experience?” by S. Khan and M. Iqbal, 2020, 2020 8th International Conference on Reliability, Infocom Technologies and Optimization (Trends and Future Directions) (ICRITO), p. 593.

Changing Tidio’s business model

The customer service software market is highly competitive, with established players like Zendesk and Freshdesk. However, the industry is experiencing significant growth, fuelled by AI advancements. Tidio’s HILAI is well-positioned to capitalize on this growth by addressing the limitations of existing chatbots and offering a more human-centric AI solution. However, to improve the outcome we suggest some business model changes such as partnerships with companies like Cursor.sh and LLM providers to leverage expertise and resources and an increased focus on customer support and marketing to educate customers about HILAI. 

Conclusion

Although Tidio faces technical and organizational challenges in implementing this product, it represents a significant step forward in customer service. By combining the power of AI with the irreplaceable human touch, Tidio will redefine the customer service experience and strengthen its market position.

No ratings yet.

My personality is absolutely not a great fit for coding. I do not have the patience and if I get error messages because I forget a ‘,’ somewhere, it drives me mad. However, Generative AI tools have been great to learn and debug coding. I had some basic experience with Javascript but I never used Python before. So when I needed to rewrite Javascript in Python, I used Google Gemini to help me do it. It was a great tool to use and if I did not understand something, it would give me examples and explanations. After this, I was able to quickly learn the basics of Python and increase my coding skills.

Currently, I use it often when I have to code something or debug code. Although it does not always work greatly, it usually brings me in the right direction and gives me tips to improve my coding techniques. Sometimes, when it gets a bit more complicated with different frameworks and APIs, the AI makes mistakes. In those cases, it is unfortunate that it not just tells me that it is not able to solve it but gives a random imaginary coding sample that is useless. In those rare cases, it takes actually more time to use AI tools compared to not using them.

Before, I needed to search the internet for the specific use cases and errors I was facing when something went wrong. And if I could not find it, I had to use Stackoverflow to help me out. However, waiting for an answer takes time and in the meantime, I could not make any progress. AI makes that progress much faster and is able to give personalized responses.

In conclusion, I believe that AI has made coding much easier and even helps people who, like me, do not have the patience to find tiny little mistakes.