Finally — Actually Useful AI in Industrial Automation

TL;DR: FlowFuse 2.3.0 introduces FlowFuse Expert, based on natural language processing, allowing users to natively build industrial application workflows and dashboards through conversation, generate complex factory simulation data, and integrate with NCP servers for advanced data analysis. ## Natural Language Workflow Building with FlowFuse 2.3.0 and FlowFuse Expert Version 2.3.0 of FlowFuse is now available. Christopher Sandoval, Developer Relations Advocate for FlowFuse, points out that the most prominent feature in this update is the major upgrade to FlowFuse Expert. Leveraging natural language processing technology, users can now natively create industrial applications, fundamentally transforming the way workflows are developed. ### From Prompt to Full Workflow: A Weather Data Example To demonstrate this feature, let’s take the example of creating a weather data workflow in a new FlowFuse instance. Users simply need to paste a specific weather API URL into the chat window and ask FlowFuse Expert to generate the corresponding workflow. The system automatically builds the workflow within seconds. Notably, FlowFuse Expert not only invokes the necessary nodes but also automatically connects them. This means that with just one prompt, users can obtain a fully functional end-to-end workflow. ### Smart Dashboard Generation and Dependency Management Beyond basic workflows, users can also ask FlowFuse Expert to build dashboards. For instance, they can request that the output of the "Get Weather Data" node be rendered as text in a specific format. During this process, FlowFuse Expert demonstrates its intelligent detection capabilities: 1. **Dependency Detection**: If a node required for the dashboard is not installed, the system detects this missing dependency and prompts the user to install it by opening the Palette Manager. 2. **Context Memory**: Once the user confirms installation is complete and tells the system, "Installation complete, please proceed," FlowFuse Expert reviews the conversation history, extracts the original request, and generates the required dashboard within seconds. ### Iterative Optimization and Error Fixing In practical applications, generated dashboards may need adjustments. For example, temperatures are displayed in Celsius by default, but might need to be changed to Fahrenheit for users in the United States. Users can directly request FlowFuse Expert to change the unit. Furthermore, FlowFuse Expert can assist in fixing logical errors: * **Location Identification**: Even without additional context, if a user asks for the geographic location corresponding to specific latitude and longitude coordinates, Expert can accurately answer and help add text elements to display this information. * **Error Diagnosis**: If duplicate labels or logical errors are discovered during deployment (such as incorrectly copying a location label instead of displaying the site name and administrator name), users only need to describe the expected behavior. FlowFuse Expert will inspect the created content, locate the root cause of the problem, and automatically apply fixes. * **Label Correction**: For incorrect label naming, users can point this out directly, and Expert will instantly correct these metadata items. ### Complex Industrial Simulation: From Single Machine to Factory-Level Simulation The capabilities of FlowFuse Expert extend beyond simple data display; it can also generate complex simulated devices and workflows. **Single Machine Simulation** The user requests the creation of a workflow simulating a factory machine named "Hanover press." * **Configuration**: Set the timer to trigger repeatedly, with an initial interval of 1 hour (which can later be adjusted to every 5 seconds to speed up the demonstration). * **Result**: After deployment, the machine begins generating all requested data points. **Large-Scale Factory Simulation** To demonstrate scalability, the user requests a factory simulator containing 50 machines. * **Data Points**: Each machine independently generates simulated data, including heat, kilowatt-hour consumption, vibration, number of production units, and total uptime in seconds. * **Workflow Context Storage**: The user requests that the data for each machine be stored in the flow context, named individually from `flow.machine1` to `flow.machine50`. Although this is a vague request (the specific naming pattern for the last machine was not explicitly specified), FlowFuse Expert successfully renders and deploys all data sets, ensuring data is available for subsequent processing. ### Integrating with NCP Server for Advanced Insight Analysis After storing the data for 50 machines in the flow context, FlowFuse Expert can connect to an NCP (Node-Commander/Control? Note: The original text says NCP, usually referring to a specific data analysis or control server/tool) server for deeper insight analysis. 1. **Global Overview**: The user requests Expert to retrieve information about all machines in the flow context and analyze them using NCP tools. The system quickly provides a general operational status overview of all 50 connected machines. 2. **Specific Performance Queries**: The user further asks which machines perform the best and worst regarding heat metrics across the entire dataset. FlowFuse Expert almost immediately provides exact answers via the NCP solution. ### Other Important Updates and Improvements In addition to core enhancements in FlowFuse Expert, FlowFuse 2.3.0 includes several improvements aimed at enhancing the developer experience and system stability: * **Smoother Iteration Experience**: Interaction with FlowFuse Expert is more seamless. * **Snapshot Comparison**: Improved ease of use for the snapshot comparison feature. * **Device Editor Auto-Recovery**: Enhanced reliability of the editor. * **UI/UX Optimizations**: Cleaned up tooltips and improved the badge for Git integration features. * **Enhanced Error Handling**: * Frontend chat request timeouts in FlowFuse Expert are handled more clearly. * Automatic logs now include more detailed records of stop events, helping users quickly locate the root cause of issues. ## Summary FlowFuse 2.3.0 achieves a leap from natural language to complex industrial application generation through FlowFuse Expert. Whether building simple weather monitoring dashboards or simulating a factory with 50 devices and integrating advanced data analysis, this tool demonstrates high efficiency and problem-solving capability. If you are interested in FlowFuse, you can visit flowfuse.com to start a 14-day free trial. It is recommended to review the full changelog to understand all minor fixes and new features in this update. Source: Finally - actually useful AI in the industrial automation space (https://youtu.be/DR9OTIVtBLU)