The Situation
As Alexa+ adopted agent-led development, developers were asked to work through a growing set of disconnected tools and emerging automation agents. Despite powerful capabilities under the hood, there was no single, trusted place to manage end-to-end workflows across creation, launch, and monitoring. The problem wasn’t a lack of automation. It was making a complex, multi-agent system feel cohesive, dependable, and usable across web and technical tools.
The Strategy
I reframed the problem from building multiple agent-specific UIs to establishing trust and continuity. Developers needed automation, but didn’t need to choose sub-agents of how that was done. They needed a single, reliable collaborator that could persist across the tools they already use.
I focused on a single source of truth: an agent orchestrator that unified multiple sub-agents into one coherent, user-facing presence across web, IDE, and CLI. Developers could interact with it in natural language while maintaining context as work shifted between surfaces, reducing task switching, errors, and cognitive load.
Key decisions centered on where orchestration lived and what context traveled. Surface support was prioritized based on real engineering workflows, and context handoffs were intentionally scoped by role and recipient. I also pushed for natural-language interaction and cross-platform context persistence as platform-level capabilities, while pushing back on fragmented dashboards and tool-specific agent experiences.
The Result
- Influenced roadmap and platform direction across web, IDE, and CLI
- Established natural language as a shared interaction model
- Provided a reusable UX framework for agent-based developer tooling
- Unblocked early product and technical decisions
- Raised the quality bar for agent instantiation and cross-platform DX
Amara Hulslander – UX Designer 