BambooHR is a leading human resources platform with tools for people management, payroll, and benefits. The company aims to eliminate the pain of tedious processes and workflows, so HR organizations can focus on more important things, like supporting their people and growing their businesses. And BambooHR is doing just that.

Alan Whitaker is leading the AI charge at BambooHR. Under his leadership, the company is using AI to build products that streamline operations and foster meaningful workplace connections for BambooHR customers. Growing up fascinated by computing and creative arts, Alan co-founded several startups before joining BambooHR in 2020. At the time, the field of generative AI in HR was yet to emerge, and his role involved pioneering data engineering, machine learning, and AI-powered applications at the company.

Fast forward to today, and Alan is setting the standard for secure enterprise AI adoption. “Working at BambooHR has been an incredible opportunity to harness data and algorithms to uplift both employees and employers,” Alan shared. He believes we can build more soulful technology, which can “magnify the best in humanity and help us just achieve and experience more.”

In this interview, Alan discusses the intersection of AI and HR, explores the transformative potential of conversational AI tools, and shares his vision for human-centered AI. He provides a behind-the-scenes look at the development of BambooHR’s innovative features and offers advice for businesses exploring AI adoption below.

Watch the videos or read on.

0:00

/0:45

Empowering HR with AI: Balancing people and machines

Why is AI so impactful in HR?

AI and HR are a great combination. HR is really about working with people to achieve more, but there are many processes involved. So, how can we reduce the mundane parts, allowing you to do what you do best and focus on people?

We're only beginning to understand AI's impact on HR. The Society for Human Resource Management (SHRM) reports that 61% of HR professionals are optimistic about the potential of AI in their organizations. I think the most important thing about AI for HR is this paradox: The machine can create more space for human connection and more power for performance.

We also saw in the SHRM study that 75% of respondents agree advancements in AI will increase the importance of human intelligence in the workplace over the next five years.

I think employee experience is key to creating great places to work. AI can help employers listen to feedback from their employees, tailor personalized guidance and learning for employee growth, and help people find the places where they can add the most distinct value and then move efficiently through the rest. Our customers are looking for ways to be more efficient.

0:00

/0:59

AI in HR: Reducing drudgery and enhancing human connection

What are the main concerns about using AI in HR?

HR is a high-stakes space where livelihoods are affected. We need to ensure the responsible application of any technology—particularly AI.

We must consider biases. Large language models have been shown to reflect whatever inherent biases may be in what they’re trained with. But there's a flip side to that—can it actually help complement our blind spots? For example, could AI help us look at resumes without triggering certain biases by neutralizing that identity?

There is, of course, a lot of interest in regulating this space, and various jurisdictions have already introduced regulations. For example, New York City introduced a law where companies deploying AI solutions need to demonstrate that it doesn't disadvantage certain groups of people. In the European Union (EU), many different regulations are emerging where you can't use AI for certain types of decisions or activities around HR. For example, you can't use it to monitor the emotions of employees.

I think we need to look at the intent and implementation. AI advisor Cassie Kozyrkov has a good way of talking about this. When we look at a system, what is the system for? What data does it use? And how do we know it works? Because these systems ultimately amplify the minds of the designers.

We need to make sure we're looking at different angles, carefully considering how it works, and making sure it's going to operate in an environment where it can be fair and we're not taking away human judgment from the equation.

For more on how BambooHR builds with Cohere, check out the full customer story: How BambooHR optimizes for AI search and retrieval.

What inspired the launch of Ask BambooHR, and what sets it apart?

A well-designed process and use of AI should always elevate the human. One of our AI principles is “empowerment, not replacement”. We want to make sure we're putting people in positions to do what they do best. And of course, as an HR professional, human judgment, evaluation, and connection are always going to be something we want to put at the center.

We're very thoughtful about how we balance that and make work better in this combination of people and machines.

As generative AI was emerging, we’d already been building towards wanting to have an intelligence underlying things.

We've launched Ask BambooHR, which is a conversational interface built into the platform. You can click the Ask button and have a natural conversation informed by your company’s documentation and current HR information. You’re able to ask a policy question and receive an answer about how your company deals with a specific situation, like how to get my gym membership reimbursed. Or maybe you have a question about HR information, for example, “how much time off will I have by a certain date?” Users love getting an answer immediately, and HR professionals love not answering the same questions again and again.

We're very excited about how our customers and their employees can now interact with software through natural conversation.

BambooHR optimizes AI search and retrieval

What were some of the challenges of building a RAG system for Ask BambooHR?

Making sure information was going to be accurate and kept private and secure were some of the top challenges.

Large language models are trained on public knowledge to generate language that sounds natural. We needed to make sure that we had a reliable and secure bridge to HR data that could power an accurate response. To paraphrase Cohere’s Jay Alamar, don't think of LLMs or foundational models as standalone minds. Think of them as part of an intelligent system. And it takes a lot of different parts interacting in the right ways to come up with the right answers.

In our system, multiple LLMs and all kinds of other logic are all orchestrated to form a sound system that can cross-check itself. And everyone's still learning more about that. There are lots of interesting technical challenges as we assemble these kinds of solutions.

One of the challenges is making sure that we can retrieve and provide the most accurate information. We wanted to avoid AI hallucination, so we used techniques like retrieval-augmented generation, where we go and find a specific relevant piece and inject that as context into the prompt before we generate an answer.

We wanted to ground the model to facts that are relevant to the situation. In our case, we're accessing structured HR data that's part of the HR information system and it needs to be delivered in a way that respects all of the same permissions.

0:00

/2:30

Ensuring accuracy and security in LLM-based systems

Why did you choose Cohere AI search models for Ask BambooHR?

Finding the correct knowledge to ground your answers in is an art. Without a well-considered retrieval approach, you could be answering the wrong question or you could be making things up.

We actually built some internal applications before building Ask BambooHR to help us go and retrieve pieces of customer feedback from all the different sources of interaction that we have with them. We found Cohere’s large language models (Embed 3) to be very sound and useful.

Cohere has also been a pioneer in re-ranking. And so after you retrieve a set of candidates, chunks of documents to consider, there's an opportunity to take a second pass and really assess which ones are pertinent to the question at hand. We found Cohere Rerank to be an effective way to help us be smarter with context.

Assessing quality is another key technical challenge in these kinds of systems. We could see that Cohere was great for specific parts of that puzzle, and that's been really helpful.

0:00

/1:05

Alan Whitaker on Cohere: Building with AI search models

How do you see AI evolving in HR over the next five years?

In five years, we believe HR will be able to focus more on humans and less on tedious processes, boosted by capable AI assistance and workflows. Properly built AI can become a very personal and trusted guide in your work. And there are opportunities to integrate more HR-related data and services from the ecosystem.

There are many different providers of tools and information, and can we pull more of that together and build an intelligent employee graph with connected agents helping you achieve things across that portfolio of tools. For growing businesses, the HR function will be able to achieve much more and be more strategic and human-centered.

AI will be an effective assistant to HR pros, accelerating workflows and analytics and improving efficiency. Businesses that create better employee experiences will stand out more with this AI as an accelerator.

0:00

/1:11

HR's future with AI

What advice would you give to businesses starting their AI journey?

It starts with creating an intentional space to learn and experiment so you can figure out how to put together some of these pieces.

You then need to work across your company to improve both AI literacy and access to tools so you can engage your whole workforce in this effort. If we thought of our AI team as the only place to deliver AI, we would be really limited. Instead, we see people throughout our organization taking advantage of AI and having discussions. We've worked to create that environment, set policies, and involve everyone.

And lastly, be intentional about your AI principles and how you will or won't use it. It’s important to integrate those touchstones as the company considers what to do and individual employees think about how they can apply these principles in their daily work. Having clear guidelines will help guide both company decisions and individual actions, ensuring that AI is used ethically and effectively.

AI has the chance to impact all aspects of business and life because it's the most approachable computer interface yet, drawing from the broadest base of human knowledge. I'm excited about what is possible as we aim for technology to lift us and have a more soulful approach.

About Alan Whitaker

Alan is BambooHR's Head of AI. He joined BambooHR in 2020, adding his dedication to developing technology with soul to the company’s ambition to create HR software with heart. After earning his bachelor’s degree in computer science from Brigham Young University, Alan worked his way up through technical and business roles to become the CTO of Affiliated Computer Services, a Fortune 500 company providing business process outsourcing. He then went on to found several startups focused on AI, behavioral science, data visualization, and digital marketing, working towards his dream of a symbiotic relationship between technology and people.

Unlock the potential of private and secure AI.

Get Started!

BambooHR takes on AI: A chat with Alan Whitaker