In today's ever-changing business environment, customer expectations continue to rise, making a customer-centric strategy a key success factor. Organizations recognize the importance of customer feedback and are moving quickly to meet customer expectations to drive sustainable business growth. Voice of Customer (VoC) services, which are key to improving the customer experience, are becoming more sophisticated, especially with the advent of new technologies such as AI, machine learning, and natural language processing.

Recognizing the potential of the VoC market, software as a service startup Syncly provides a customer feedback analytics solution powered by generative AI. Founded by serial entrepreneurs who successfully launched and exited their previous AI startup SUALAB, Syncly has received seed investment and mentorship from Y Combinator, and is now expanding its business globally beyond the US and Korean markets.

Customer-facing teams from companies around the world use Syncly to manage VoC data collected in real-time from chats, calls, and meetings on an integrated platform. Syncly’s AI then analyzes the VoC data to derive improvements and find solutions to enhance customer relationships.

Syncly's core service is to provide visibility into VoC by automatically processing data generated during the process of collecting, analyzing, and sharing customer feedback with AI. Beyond quantitative analytics, Syncly supports qualitative analytics through semantic search, making it critical to understand the semantic values and similarities between feedback.

In the early days of Syncly's business, there were few full-text search capabilities in the traditional SQL and NoSQL domains, and it was complicated to build a separate library-based search engine. In addition, full-text search using specific keywords or phrases was limited in its ability to take advantage of embedding vectors, the output of semantic analytics.

To address this challenge, Syncly uses AI to enable effective similarity analytics even in a mix of structured data and unstructured data such as images and videos.

"In the beginning, we loaded all the embedding vectors on the server and manually calculated the similarity," said Jongsoo Keum, Cofounder and Research Lead at Syncly. "However, as the company grew, so did the number of customers and the amount of data for analysis. The development team faced new productivity challenges as the server and database workload became heavier."

For organizations that need to understand and analyze thousands to millions of data points to find meaningful results, technology that can help reduce the time to do this is essential.

Faced with a heavy workload, Syncly proactively adopted MongoDB Atlas Vector Search, launched in 2023, and applied it to all of its vector data management services. The team's primary use of vector data is in cluster analytics, which categorizes data based on topics of customer interest and examines whether feedback is relevant to those topics.

"Storing vector data in MongoDB Atlas allowed us to focus on developing the data collection and classification services that we had committed to. We decided to use Atlas Vector Search to improve the efficiency of our data analysis," said Jin Kyoung Kim, Tech Lead at Syncly. "When there was a product update, we received active support and guidance from the MongoDB Korea team, so we were able to quickly learn about MongoDB's new features and leverage them in developing our services."

With Atlas Vector Search, Syncly improves developer efficiency and achieves significant performance gains in Syncly.

First, by loading the existing embedding data into the backend, Atlas automated the tasks of manually scanning and writing code logic, dramatically reducing the developer workload. Second, Atlas Vector Search's indexing feature automated the process of similarity analysis between data, saving server and system power, I/O and network traffic while significantly improving developer productivity.

"If we had to go through more than 10,000 similarity comparisons for a feedback analysis, we can now get the desired results with just one query by setting up a Vector Search index. We are experiencing more than 10 times faster performance with Atlas Vector Search index compared to manual processing," said Jongsoo Keum. "The simplified logic processing pipeline has cut costs and speed to market. As a startup, the lower server load has reduced our need for cloud services that require higher-tier server instances, which is a huge benefit."

In addition, Syncly could use Atlas Vector Search and Atlas Search together on the existing Atlas architecture without any additional deployment or management costs, and could take advantage of semantic search, which moves away from traditional database search and responds to advances in language models.

In a survey conducted by Retool, a tool-building platform for developers, Atlas Vector Search received the highest Net Promoter Score (NPS) and was named one of the most widely used vector databases with AI, making it an innovative solution for developing modern AI-powered applications.

"Our team has seen improvements in both customer convenience and satisfaction by delivering insightful analytics to our customers faster and improving the efficiency of our services," said Jin Kyoung Kim. "Another key benefit of Atlas Vector Search is that it is built on MongoDB Atlas, which allows us to store and manage both document and vector data in a unified platform."

As Syncly expands its service, the team is looking to incorporate MongoDB's business automation and visualization tools, as well as additional search nodes and global clustering to reduce workloads.

The team is also actively leveraging the potential of AI to develop VoC services, and is working with the MongoDB Korea team to further optimize MongoDB's products and services to make its service more secure.

"Syncly is currently working on launching a search service feature, and using Atlas Vector Search and Atlas as the core technology," said Jongsoo Keum. "We are also exploring further application of Atlas Vector Search, as we are considering using vector data as meta-information for feedback analysis in the future."

"Companies need to build products that customers need and to use technology to better understand their needs and pain points. Syncly will continue to team up with MongoDB to deliver AI solutions that effectively solve customers' problems and grow with them," said Jin Kyoung Kim.

Jongsoo Keum, Co-founder and Research Lead at Syncly

Jin Kyoung Kim, Tech Lead at Syncly