Exploring COR’s business processes

No ratings yet. One of the most valuable aspects of this project for me was being able to use an Argentinian company as our case study. I proposed working with COR, mainly for two reasons. First, I had been closely following the company for a couple of years and knew that it was very tech driven, so I thought they would be open to potentially implementing our proposed solution. Second, I knew we would be able to get access to real company information, which I considered especially valuable for the project.

I’m an exchange student. Through my MBA program back in Argentina, I met someone who had been working at COR since its early days. This connection helped our team obtain a real data set, including specifics on COR’s business, operational challenges, and costs.

I think our proposal would have been much more difficult to implement in practice (and less realistic) if we hadn’t had access to this information. Understanding COR’s existing systems and technologies was key to propose a solution that is relatively easily to integrate to their current infrastructure (rather than making assumptions around how they work).

More specifically, COR provided us with information about the costs associated with the employees currently handling customer inquiries, which was particularly valuable in determining whether our proposed solution would make sense from a financial perspective. We also learned about their existing customer support process, where employees obtain the information they need to answer questions (FAQs, previous emails and other historical knowledge), and which systems the company currently uses.

My biggest learning, however, relates to the actual implementation of the solution. I don’t come from a technical background, so I initially didn’t realize that implementing an AI agent would require an entire ecosystem of connected technologies. Through discussions with my teammates, who have stronger technical backgrounds, I learned that we would need tools such as n8n to orchestrate the workflows, Jira as a ticketing system, and PostgreSQL as a database, in addition to the LLM itself. Initially, I thought we could simply integrate an LLM into the company’s existing processes, but I realized that the implementation was more complex than I had expected, although still feasible and cost efficient.

As a team, our biggest back and forth was around the scope of the solution and the pilot. We were initially a little too ambitious and wanted the AI agent to respond to all customer inquiries.

However, as we discussed the implementation in more detail, we realized that this approach was overly broad and could expose the company to unnecessary risks, particularly if the AI provided inaccurate responses or failed to handle more complex inquiries appropriately. Consequently, we decided to narrow the initial scope to specific product-related questions, where the AI could rely on COR’s existing documentation. Specifically, we thought that, at least in the beginning, the AI could handle questions on whether the client’s specific plan covered X or Y functionality, the features of other plans and help with other minor tasks as scheduling demos.

While the scope is quite narrow, I do think this could lead to a significant impact. Today, COR employees that handle customer inquiries take 1 to 2 business days to respond to a customer inquiry, whereas with our proposed solution customers could get answers within seconds. This could ultimately lead to less friction in the customer experience and more sales as the customers get smarter and faster responses on what other COR products do.

We also discussed the importance of testing the solution before implementing it more broadly. Rather than launching the agent immediately, we proposed starting with a pilot to evaluate the accuracy of responses, response times, and ability to identify questions that require human assistance. This would allow COR to identify potential issues and make adjustments before expanding the scope of the solution (which is the ideal future state).

Another very valuable aspect of this project was having the opportunity to evaluate a business problem from a different perspective. As an attorney, I am used to receiving a solution that has already been developed by the business and being asked to assess the associated legal risks. This project was different because I had the opportunity to be involved in designing the solution from the beginning, deciding what to implement and how to approach it, in addition to considering the potential risks. I found this particularly refreshing and enjoyable, as it allowed me to be much more creative than I usually am in my day to day work. It was interesting to be on the other side of the decision-making process and to think not only about what could go wrong, but also about what could be built and how it could create value for the company.

Overall, it was a very enriching experience as I had the opportunity to work with teammates from different professional backgrounds. I learned a lot from my teammates about the technical aspects of implementing AI.

I think in the end we were able to see in practice that transformation is not about just implementing technology but about profoundly understanding the business challenges and thinking of a solution that is tailored to that business and those specific customers.