AI Voice Bot
About
Voiso is a CCaaS platform with an AI layer spanning speech analytics and conversational automation. This project: a self-service experience for building and running AI voice agents inside the platform. I designed the configuration experience from zero.
Task
An AI agent configurator lives on a hard tradeoff: expose real power, a prompt, a voice, knowledge, escalation, in a form a non-technical contact center admin can operate without an AI engineer in the loop. My mandate was to design that self-service path end to end: an admin builds an AI agent, teaches it from their own documents, verifies its behavior, and puts it on a live queue on their own.
Solution
I led the project, and we built it as a pair: the domain designer and I. The designer ran the research and proposed the first concepts; together we made sure the design met the functional initial requirements without being confined by them, so the architecture could scale from day one. The clearest example is the core architectural call I drove: knowledge bases as a separate first-class entity rather than uploads buried inside an agent. That split future-proofs the product in both directions: any complex configurator or builder can grow inside knowledge bases without redesigning the agent, and the same knowledge base can be given to any agent, so one brain serves many agents.
Business Impact
- Enabled customers to deploy AI-powered voice experiences without specialist technical knowledge.
- Reduced implementation complexity for conversational automation use cases.
- Expanded Voiso's AI product offering into a rapidly growing market segment.
- Improved accessibility of AI capabilities for existing customers.
The full case study, including the design decisions and the whole process in detail, is something I'd be glad to walk through on a call.
