VulcanAI: tools, plans and the new Textualize console

As AI systems become part of real robotics workflows, one requirement is becoming impossible to ignore: transparency.

It is no longer enough for an assistant to return a final answer or execute a command successfully. Teams need to understand how that result was produced, which tools were selected, what sequence was followed and where things failed when behavior is not as expected. VulcanAI is built around that practical need. It combines planning, reasoning and tool orchestration in a way that keeps execution inspectable, making it easier for developers to trust what the system is doing and improve it over time.

At the center of that design is a tool-based architecture. In VulcanAI, capabilities are exposed as tools instead of being hidden in opaque prompts, so agents can call concrete operations in a structured manner. This includes custom tools and default ROS 2 tools that wrap common operational commands for nodes, topics, services, actions, parameters and interfaces. From a product perspective, this matters because it turns AI behavior into something modular and extensible. Teams can add new capabilities incrementally, reuse existing integrations and keep reasoning logic separate from execution details, reducing friction as applications evolve from prototype to production workflows.

Planning is treated with the same level of structure. Rather than relying on untracked reasoning, VulcanAI represents plans through explicit elements such as `GlobalPlan`, `PlanNode` and `Step`, making execution intent visible before and during runtime. Outputs from one step can be reused by others through blackboard references, and iterative execution can verify progress, avoid repeated plans and adapt when a path does not work. This creates a controllable loop between intent and action.

That is where the new Textualize console becomes especially relevant. Instead of a minimal terminal experience, VulcanAI provides a more visual and understandable interface for inspecting tools, plan history and execution outcomes. Commands such as `/tools`, `/plan`, `/history` and `/rerun` make it easier to see system state, revisit decisions and debug behavior without leaving the workflow context. The console also centralizes logs and execution feedback, helping teams understand not only what happened, but why it happened. 

Inside VulcanAI, the key idea is simple but powerful: smarter automation needs understandable automation. Tool-based design gives agents practical reach, structured planning gives them direction and the Textualize console gives developers visibility into every step. Together, these pieces position VulcanAI as a product layer for teams that care about both capability and control. As robotics applications become more adaptive and goal-driven, that balance will define which AI systems remain manageable at scale and which become black boxes. 

Learn how to install VulcanAI here!