Agentic ChatBotOur client with 25,000 users needed a way to collect complex, highly variable information but the traditional approach wasn't working. Users ranged widely in their understanding of the process, and the data required was anything but straightforward. Missing information triggered follow-up questions, certain answers opened entirely new branches of inquiry, and the whole thing needed to feel effortless regardless of who was using it. On top of that, all of the collected data had to flow into a single dashboard where safety personnel could quickly assess conditions and adjust processes to keep people safe in high-risk environments.
The Approach
We went straight to the source. We interviewed users at client companies, visited job sites to observe how people actually filled out forms in the field, and ran usability testing with our clients to validate every assumption. What we found confirmed what the research already told us that linear, question-by-question forms don't match how people naturally communicate. So we built a non-linear, AI-driven input experience. Users simply talk into their phone, sharing whatever they know in whatever order it comes to them. The system analyzes the response in real time, identifies gaps, and prompts only the follow-up questions needed to complete the submission. Users could also type or fill out the form manually. On the back end, dynamic branching logic and AI processing ran with near-zero latency, keeping the experience fast and invisible.
The Outcome
The tool leveled the playing field. Users across all experience levels completed submissions at nearly the same speed, regardless of how much they knew going in. Voice, text, or manual entry, every path led to the same complete, high-quality data. That data fed directly into an intuitive, role-based dashboard that gave safety teams the visibility they needed to act fast and keep people safe in real time.