As federal agencies adopt artificial intelligence (AI) to modernize services, the stakes are higher than ever. 92% of executives expect to increase AI spending in the next three years. But history offers a warning: 70% of major transformations fail, not because of technology, but because organizations struggle to align tools with people, processes, and mission needs [McKinsey and Company 2019; 2025].

The government’s America’s AI Action Plan underscores this reality, urging agencies to adopt AI in ways that are secure, practical, and people-centered. Platforms like USAi.gov now provide agencies with secure access to leading AI capabilities, but tools alone aren’t the answer.

The real challenge lies in adoption. While agencies experiment with chatbots and other AI tools, success depends on whether those tools feel transparent, accurate, and accessible to the people who rely on them. 

In the government, that means human-centered design, so tools are not just technically robust but genuinely useful and easy to engage with. 

Reimagining Public Service with Generative AI Chatbots 

When we talk about GenAI-driven chatbots in the context of government services, we should design them to support users along their unique journeys. Consider, for example: 

• A teenager applying for a driver’s license goes through the challenges of overwhelming paperwork, new rules to learn, and the pressure of passing the written test.
• A researcher looking for government datasets to complete their data analysis, running into outdated or incomplete datasets, and confusing documentation.
• A Medicaid beneficiary navigating the complexities of enrolling in the right plan for their needs, often facing confusing eligibility requirements, and uncertainty about what benefits are covered. 

In each of these scenarios, an AI assistant could help. But users are not necessarily searching for the fastest possible outcome. What they crave is a sense of comfort, a space where questions are welcomed and mistakes are gently avoided. 

People want to feel reassured that the technology is there to support—not replace—their human agency. The true promise of AI in these public-facing environments is its ability to transform exhausting bureaucratic interactions into approachable and trustable experiences. 

In government services, success is measured not by speed, but by users getting the right support when they need it. 

The Story of Bixal’s AI Journey — and What It Means for Federal AI Adoption 

Our journey began the way many federal AI initiatives do, with a bold vision and an eagerness to explore new possibilities. We gathered as a team to identify the everyday challenges our peers face.  

Together, we brainstormed how AI could:

• Streamline workflows
• Improve efficiency
• Add measurable value across operations

Energy and collaboration fueled our early efforts, leading us from Miro boards to fully realized prototypes. 

First Results: Not What We Expected 

Our initial AI prototypes generated curiosity and interest across the organization. Team members at all levels—individual contributors, managers, and senior leadership—responded positively to our demonstrations, and feedback during these sessions was encouraging.  

But when it came to day-to-day use, adoption was low. Even Bixal AI, our own web-based AI chatbot powered by a homegrown language model, saw limited use and remained underutilized.   

Internal surveys confirmed what many federal programs discover: curiosity doesn’t automatically translate into use. 

This is a familiar challenge for agencies, where new tools launch with excitement, only to see minimal adoption because they don’t fit seamlessly into established workflows or address user concerns upfront. 

Integrating UX Research in Our Product Development Process 

We knew we had to dig deeper. Our product manager and a dedicated user experience (UX) research team joined the project, conducting structured surveys, interviews, and moderated usability tests. The insights were eye-opening: 

• We had assumed most staff would know how to start using the tool—many didn't due to lack of guidance.
• Advanced features weren’t intuitive without training.
• Writing effective prompts was challenging.
• Concerns about responsible AI use were front of mind. 

At the same time, we heard strong interest in AI’s potential for automating routine tasks, improving data analysis and content development support, and enhancing strategic decision-making. This created a sense of cautious optimism about the future role of AI at Bixal. 

For agencies, this reveals an important truth. Technical capability is only half the battle: Trust, usability, and readiness drive adoption. 

The Turning Point: Adoption Through Connection 

With this new understanding in hand, we integrated the ongoing user feedback loop to refine the Bixal AI chatbot to be simpler, clearer, and better aligned with daily workflows.  

Interestingly, the biggest increase in usage didn’t come after a major software update, it came after a Community of Practice session. In that session, we walked colleagues through the features and showed exactly how the AI chatbot could help them in their daily work.  

Usage spiked because we stayed connected, listened, and made adjustments based on real user feedback. 

For agencies, this can mean: 

• Hosting live demos for program teams and field offices.
• Offering hands-on training tailored to real work scenarios.
• Building trust by explaining safeguards and addressing concerns directly. 

What Federal Programs Can Take Away 

Our AI journey confirmed one thing: Successful AI adoption is a people-first challenge, not a technology-first one. 

By listening early, designing with real workflows in mind, and investing in ongoing engagement, agencies can ensure their AI tools don’t just launch—they deliver lasting mission impact. 

Everything we learned during that experience continues to inform how we develop our AI products today. The journey is ongoing and grounded in genuine user interactions, feedback, and engagement efforts. 

Final Thought: The Value of a People-First Approach 

At its core, the evolution of AI and UX is an ongoing partnership between technology and those it serves. We have seen that the greatest leaps forward in AI adoption come from persistent engagement—listening closely to users, iterating, and demonstrating real-world value in person. 

In an age of rapid innovation, the people-first approach ensures that technology remains a tool for empowerment, connection, and meaningful progress, making each breakthrough not just impressive, but truly transformative in the lives of those who use it. 

Ready to Bring AI from Pilots to Production? 

Bixal helps federal agencies design AI products people actually use—built on trust, transparency, and measurable results. Contact us today by submitting the form below.