How long did it take from concept (Maslow Insights AI tool) to go live?
Background:
Last summer (2024), our interns developed two interactive chatbots: an “Industry Expert” and an “Xceleration Expert.” The success of these projects inspired us to productize these personas for both internal and external use, from engaging with prospective clients to onboarding new employees.
This initiative also gave us the blueprint for Maslow Insights, a dual-purpose B2B employee recognition expert. Maslow Insights is built with over 25 years of industry and client solutions experience, providing it with deep contextual knowledge.
Throughout the fall and early spring, we designed an efficient architecture and created a roadmap to support two primary use cases: public industry interactions and client-specific interactions.
We launched a public-facing pilot for our employees in March 2025. In April, we released our first product feature in RewardStation for participants to use. Once the infrastructure was established, we were able to move from concept to production in approximately two sprints.
What guidelines were put into place?
Our customers are multi-national companies across multiple industries. As a result, our guiding principles around security and data privacy remained the same.
Sample:
Your Data Stays Yours
· We use pre-trained language models – we do not train or fine-tune models on customer data
· Customer data is not retained by AI or used to improve models
· All data transmission uses industry-standard encryption
Enterprise Security
· Built on the same secure infrastructure trusted by Fortune 500 companies
· SOC 2 and ISO 27001 compliant platforms
· Audit trails for all AI interactions
· No new security vulnerabilities – operates within existing secure cloud frameworks
· AI features remain secured behind the same RewardStation authentication and access controls
Note: We consider the project details and contextual knowledge that is used to produce the desired outcomes proprietary.
How did you decide what content to have Maslow ingest?
For 25 years, we’ve been building and operating enterprise-scale employee recognition programs. We’ve made the simple decision not to use customer data to train our models or produce industry benchmarks. Our algorithms are designed with strict guardrails, ensuring they only analyze customer-specific data on a client-by-client basis. For client’s interested in benchmark data, we can use the public facing expertise to provide those insights as an integrated interaction.
How do you continue to evaluate the platform?
The introduction of AI has catalyzed a cultural shift, moving us to an AI-first mentality. This change is reflected in our daily interactions and weekly sprints. We see this as an extension of our traditional software development lifecycle, but one that now requires us to prioritize AI at every step, including planning, design, implementation, and support.
What lessons did you learn?
We’ve always operated with a “test and learn, fail fast” philosophy. However, the sheer speed and volume of new tools became overwhelming. Initially, we found ourselves chasing every ‘shiny new object’. We’ve since returned to our core design principles and selected a robust tool set that we’re confident in, allowing us to evolve in sync with their release schedule.
Any interesting stories where you were surprised, something unexpected occurred?
The initial out-of-the-box insights that Maslow Insights produced on our own company’s program were fascinating. Despite our 25 years of experience, program results can sometimes be viewed through a filtered lens. The immediate insights and recommendations were fresh, including the 30-60-90-day recommendations to increase employee engagement and participant interactions.
Too often, programs focus solely on the top nominators or recipients. Maslow Insights, however, analyzed departments with low engagement and provided actionable recommendations. It also began drawing correlations between top nominators who were also top award recipients and compared different cohorts based on their tenure. Very cool insights.
What would you do differently?
Looking back, I would have prompted the AI engine from the very beginning to “hit the easy button.” While I’m half-joking, I genuinely believe it took us too long to fully embrace an AI-first culture. The lesson learned is this: if you’re uncertain about the best direction or implementation path, start by asking the AI. Then, rely on your team’s expertise and instincts to take action and build upon that foundation.