Transforming Customer Service with Generative AI

The Journey Towards Seamless Integration of Neople Assistants in Business Environments

Neople is a startup that uses generative artificial intelligence (GenAI) to create digital co-workers, called Neople Assistants, that help businesses achieve their customer service goals. Neople Assistants are uniformly integrated with the tools businesses already use to process customer support cases. In order to deliver world-class support, these digital co-workers must process large amounts of information from a variety of sources before responding to a customer case, just like a human support agent would.

The company’s brain, or operating hub, is built on AWS with Weaviate as their vector database of choice for reliable, accurate and fast querying of company-specific knowledge. With a scalable and secure AI stack, Neople’s powerful platform is setting the standard for trusted human-AI interaction and has the potential to revolutionize the way we approach work.

Challenge

As Job Nijenhuis, co-founder and CTO, began building the Neople platform, there were several key requirements for their AI stack. First, they wanted to have full control over data in the platform. In addition, they needed to use tools that were easy, quick, and reliable to set up. Neople didn’t want to build their own database solution because they wanted to spend less time on building out their stack and more on features for their Neople Assistants.

In the beginning, when Neople’s database was built with Postgres, it was impossible to provide   real-time responses to user queries. Neople Assistants search multiple times per query, and it could take minutes to respond to a Teams or Slack message, at which point a human could have completed the same task as a digital assistant.

Why Neople Chose Weaviate 

For Job, Weaviate was the obvious choice to replace Neople’s original Postgres database. As a full-featured AI-native vector database, it gave Neople the ability to have full control over how they built their AI application and how they would store their data. The database needed to run inside the Neople tenant and Weaviate was easy to deploy and manage with Docker and AWS CloudFormation templates.

Reliability and ease of use have been key for Neople as their customer base continues to grow – they wanted to focus more on improving their digital assistants and less on maintaining the underlying technology.

Although Job had initially built custom components like a re-ranking mechanism to improve the accuracy of query results, Weaviate offered a reranking module out-of-the-box. Weaviate’s powerful hybrid search capabilities promised better quality search responses while a large and active developer community could provide the support Neople’s engineers needed to quickly debug issues and focus on new features.

Solution

Weaviate replaced Neople’s initial database. In addition to being quick and easy to set up, Weaviate was purpose built for AI applications and it showed in its ability to deliver significantly faster query results. Being able to take advantage of plug-n-play modules meant Neople could drastically improve search result accuracy while reducing the developer toil of maintaining custom built components.

As an active member of Weaviate’s open source engineering community, Job and his team have been able to get quick support on challenges from Weaviate’s team as well as other community members. At the same time, he’s able to give back to the community by sharing things he’s learned along the way.

“We're really eager to learn from other people and to use services like Weaviate that specialize in vector databases to improve our offering. In the end our main goal isn’t to maintain a database, it’s to deliver the best results and user experience for our customers”
Job Nijenhuis, co-founder and CTO, Neople

Next Steps 

Neople’s technology is constantly evolving as the number of Neople Assistants working in real world virtual environments increases. Job and his team regularly adopt new Weaviate features as they become available:

“When we get a newsletter announcing that a new version of Weaviate is out, I read through it on the train back home – that night we will have half of the new features implemented. It really shows our team’s dedication to improving our offering and how choosing Weaviate continues to support those efforts on a regular basis,” said Job.

Some of Weaviate’s newest features include multi-tenancy and multi-modality, which Job’s team is incorporating into the Neople platform this year.