Optimizing Enterprise-Level HR with GenAI

Industry

Human Resources, Employee Survey Administration and Analysis, 
Business Strategy

Tech & TOOLS

GenAI, API Gateway, Lambdas and DynamoDB for Backend, AWS Batch for data and ML pipelines, Comprehend for PII removal, Bedrock Claude LLMs

Teams & Services

DevOps, ML Engineering

milestones

-- Sliced key summary task from 1,200 hours to 4.5 hours

-- Built People Science AI Platform

Industry

Human Resources, Employee Survey Administration and Analysis, 
Business Strategy

Tech & TOOLS

GenAI, API Gateway, Lambdas and DynamoDB for Backend, AWS Batch for data and ML pipelines, Comprehend for PII removal, Bedrock Claude LLMs

Teams & Services

DevOps, ML Engineering

milestones

-- Sliced key summary task from 1,200 hours to 4.5 hours

-- Built People Science AI Platform

WorkBuzz captures reams of employee feedback to help HR leaders enhance team performance and build companies where every voice is heard. These mountains of data are often too massive for humans to efficiently handle.

The situation

Among WorkBuzz’s clients are major enterprise companies that manage thousands of employees across numerous departments. WorkBuzz CTO Lewis North describes the “two-sided problem” these companies face: First, managers are often promoted into leadership positions without the experience or necessary training on how to effectively manage people. Additionally, HR departments tend to be overburdened and under-resourced, lacking the capacity to support these new leaders with the information and direction they need to understand, analyze and take action.

The challenge

WorkBuzz made it their mission to solve the two-sided problem. However, their People Science team of trained consultants wasn’t large enough to support all of their clients, so they needed a technology solution to allow them to rapidly scale. They pondered a way to use generative AI to automatically summarize massive amounts of employee-survey data points and from them generate actionable insights for HR managers. Their own internal team lacked the experience with LLMs required to build a functional solution, so they took the advice of their AWS representative and came to Loka for help.

The solution

Loka’s engineers built a pipeline that receives anonymized survey results, pre-processes them and feeds the data into an LLM. The model then generates actionable insights that follow WorkBuzz’s unique assessment guidelines and evaluation framework, then delivers them to HR staff via the WorkBuzz platform. The entire platform requires a fraction of standard people hours and compute resources. WorkBuzz calls this productionized service People Science AI.

What we delivered

The People Science AI platform processes survey data and analysis exceptionally accurately at speeds far faster than human capability, even at enterprise-sized companies. With it WorkBuzz now has a differentiator and new flagship offering to bring to clients.

Project results

People Science AI generates 1,800 survey summaries in 4.5 hours—more than 260 times faster than the ~1,200 human hours it used to take.

WorkBuzz is saving staffing costs of at least $40/hour per employee. With just one hour saved per person per day, a 100-person company sees more than $1 million in yearly savings.

Improved sentiment analysis accuracy to 91% from 75%.

Loka helped bring People Science AI to market exponentially faster than WorkBuzz could’ve done on their own, a crucial efficiency for their 2025 sales strategy.

“Loka impressed me at every stage of our collaboration. Their engineers arrived prepared to help and their project managers ran a tight ship, kept a shared Slack channel open for easy communication and delivered useful demos.”

Lewis North
CTO at WorkBuzz