Live Coding Second Breakfast: A Smart Garden Project with ServiceNow AI Agents

[Kristy Merriam]

In the latest Live Coding Second Breakfast, I introduce a new ongoing build: a Smart Garden system using ServiceNow’s agentic workflows and real-time sensor data. This is more than just a hobby project; it’s a showcase of how the AI Agent Studio can power intelligent, autonomous systems even in unconventional use cases.

The idea is simple: plants with sensors report back temperature, light, and soil moisture data. However, instead of relying on static thresholds or manually tuned scripts, the new approach utilizes agentic AI to dynamically determine what the plant needs: water, repotting, or simply some shade.

I walk through:

• Setting up a new application scope for clean separation of AI logic

• Defining a Smart Garden agentic workflow with clear steps and descriptions

• Creating a specialized "Plant Care Specialist" AI agent that uses context from the plant database and sensor readings

• Leveraging ChatGPT to enhance workflow instructions and agent roles (and learning the difference between GPT-4 and 5 outputs)

• Debugging tricky integration with MetricBase and making sure sensor data is passed correctly

• Designing reusable flow actions and tools for long-term agent memory and automation

There's also an honest look at what doesn’t work: odd naming inconsistencies between "Use Cases" and "Agentic Workflows", frustrating data returns from MetricBase, and the classic "I forgot to publish the flow" bug we all know and love.

Whether you're building a smart garden, smart office, or smart anything, this is your chance to see agentic AI in action, built live and debugged in real time.

Stay tuned. We're streaming weekly at 11AM ET!