In a recent webinar, industry experts David Stewart, Principal Solutions Architect at Cohere, and Edward Robinson, Senior Software Engineering Team Lead at Borderless AI, sat down to discuss the innovative use of large language models (LLMs) in the realm of Human Resources (HR), and their collaboration on using generative AI and retrieval-augmented generation (RAG) to build global HR solutions.

With a focus on practical applications and the use of retrieval-augmented generation (RAG) systems, the discussion centered around Borderless' AI-powered, global HR assistant, Alberni, and the technical intricacies of its development.

Borderless AI and Cohere Webinar

Here are the top five takeaways from their conversation.

1. Multilingual AI is key

“With Borderless AI, businesses can compliantly hire and manage talent worldwide without establishing a foreign company in all the jurisdictions that they're hiring in.” Edward Robinson, Borderless AI

The four-month collaboration between Cohere and Borderless AI resulted in a RAG-powered AI agent designed to simplify global HR processes. Alberni integrates generative AI to simplify HR tasks, including contract creation, vacation requests, and understanding HR laws across different jurisdictions. It can answer queries about HR laws, assist in employee onboarding with complex question-answering, and provide information about an organization's vacation schedules and policies. The tool addresses the complexities and risks associated with hiring international talent, allowing businesses to operate without establishing a local business entity in every location they hire. And Alberni does it all in 170 languages!

2. User interface matters

“Our goal is to give all of our customers the ability to do everything via AI or GUIs, providing a mix-and-match experience that simplifies HR tasks.” Edward Robinson, Borderless AI

During the webinar, Edward demonstrated Alberni’s capabilities in handling various HR tasks through an interactive and intuitive interface. The demo showcased how the AI agent can assist with general HR queries, employee onboarding, and internal company questions. Alberni’s ability to handle complex questions, such as comparing minimum wage laws across different jurisdictions, highlights its practical utility in everyday HR operations.

3. Simplicity comes first

“Less is more. By simplifying our document structure, we achieved better performance and accuracy from our RAG system.” Edward Robinson, Borderless AI

A significant technical takeaway from the webinar was the importance of document processing in the RAG architecture. Initially, the team experimented with complex, structured data formats, but they found that simpler, more human-readable text formats were more effective. By working closely with Cohere, they were able to improve the performance. By breaking documents into natural, readable chunks and avoiding overly engineered tagging systems, they improved the AI agent's ability to retrieve and generate accurate answers. This highlights the importance of designing data for human understanding.

4. RAG delivers performance

“Accuracy is paramount. We aim for 99.9% correctness in our AI responses to ensure our customers' trust and reliability.” Edward Robinson, Borderless AI

Borderless AI places a high priority on the accuracy of Alberni’s responses. The webinar hosts discussed how early iterations of the AI agent faced challenges with outdated information and hallucinations. By implementing RAG, Borderless ensured that Alberni accesses the most relevant and up-to-date documents. The team also emphasized manual testing, human assessments, and continuous evaluation to maintain high standards of accuracy and ensure the model’s reliability — crucial to building trust in HR applications.

5. Innovation is ongoing

“Web Sockets have been a game-changer, significantly enhancing the responsiveness of our AI and improving user interaction.” Edward Robinson, Borderless AI

One of the key innovations discussed was the use of Web Sockets to improve the responsiveness and interactivity of the Alberni application. By enabling real-time, token-by-token responses, the system offers a smoother user experience. Additionally, the session highlighted the use of modular conversations to manage long and complex interactions, ensuring that the tool maintains a consistent level of accuracy throughout extended dialogues.

Overall, the webinar provided valuable insights into the practical application of RAG for Enterprise AI, particularly in global HR operations. Borderless AI's Alberni, powered by Cohere’s Command R model, exemplifies how advanced generative AI can simplify complex tasks and enhance business efficiency. For those looking to streamline their HR processes, the innovations and approaches discussed in this webinar are both inspirational and practical.

Boost your RAG solution with our expert team. Contact us today to explore your options.

Get Started!

An inside look at building a RAG-powered AI agent for HR