[gtranslate]
Ascafe – Sales and Internal Chatbot

We implemented a sales chatbot on a new online shop selling coffee to individuals for a client who was previously mainly focused on B2B sales.

The goal was to make the online store more engaging for customers and increase sales. Our client, Ascafe, is a Polish family business roasting niche kinds of coffee and selling it worldwide. The main challenge was to adjust the chatbot to answer intricate questions from clients who were looking for very specific coffee qualities (roasting methods, flavors, place of origin, bitterness) and who wanted to know more about the company itself (plus standard topics like delivery and payment methods). Since the client didn’t have a single point of truth, we built a knowledge base using materials from various sources and formats (Excel sheets, PDFs, website links, Word documents, and others). Another challenge was to make it cost-effective and ready to answer questions without the limitations of mass chatbots available on the market and to prepare it for high traffic (standard bots have request limits). We have successfully fine-tuned our chatbot on a large language model with only 8 billion parameters. This makes it 50 times more cost-effective compared to models with 400 billion parameters (as more parameters generally lead to higher costs per response).This means we can deliver near GPT-level accuracy for your business, even with a high volume of requests and a limited budget. Because the old online shop was outdated, we decided to build a new one from scratch and integrate the chatbot into it. As you can see, we are not afraid of any challenge. Even if it’s not strictly within our business domain, we find a way to do it—or find someone who can. We took the end-to-end solution promise literally and very seriously, ensuring that the project was delivered as our client wished and on time, without excuses. The project ended successfully, providing our client with a new sales channel. The chatbot conversation history offers valuable real-time data about customer needs, which is also beneficial for their B2B operations. This allows them to identify the best products to sell.  Following the success of the online chatbot, the client asked us to create an app for their sales representatives to streamline B2B sales meetings. The goal was to develop a responsive knowledge base in a form of another chatbot where the sales reps could get answers within seconds without the need for additional calls or postponed meetings. We delivered this along with a dedicated web app, enabling the sales team to conduct more effective meetings and answer even the toughest questions in both Polish and English. We cannot, of course, show you the B2B sales chatbot version due to confidential information, but you can see how the online shop chatbot works by visiting: sklep.ascafe.pl (check the chatbot on our client’s online shop).