Introducing AI Pipelines

We’re excited to introduce a powerful new feature in our software: AI Pipelines. With AI Pipelines, complex real-time monitoring rules – for example in Condition Monitoring – can now be created more flexibly, intelligently, and efficiently. A Large Language Model (LLM) supports users in generating analysis logic directly as code.

The Challenge: Complex Data, Diverse Requirements

In today’s production environments, massive amounts of machine data are generated in real time across different systems. These data streams hold valuable insights for improving efficiency, quality, and equipment uptime — but analyzing them is often a real challenge.

Typical use cases include:

• Threshold monitoring of process parameters
• Data enrichment through calculations or metadata
• Force–displacement curve analysis in testing processes
• Trend and anomaly detection for early fault recognition

In all these scenarios, simple static analysis is rarely enough. Especially in Condition Monitoring, it’s crucial to detect deviations from normal operations immediately and alert the right people before a fault or quality issue occurs.

However, every industrial application is unique: Whether it’s assembly, inspection, die casting, injection molding, or extrusion — the relevant parameters, algorithms, and logic differ widely. The key challenge is to provide tools that are both flexible and powerful, yet easy to use for domain experts — enabling them to design and adapt their own analyses without deep data science or programming expertise.

Pipelines in the Bytefabrik IIoT Application Platform

To address this challenge, the Bytefabrik IIoT Application Platform already offers a powerful Pipeline Editor. Through an intuitive drag-and-drop interface, users can connect live data streams — such as temperature, pressure, or flow rate values — and design their own data processing and analysis flows in real time.

Each pipeline consists of modular building blocks, ranging from filters and mathematical operations to visualization and notification modules. With more than 100 preconfigured components, users have access to a broad toolkit — like a Lego box for industrial data analytics.

This approach combines ease of use with high flexibility. However, it also reveals the classic trade-off between simplicity and expressiveness: While our SDK allows users to create custom algorithms, many still find it challenging to select the right components or build more complex rules efficiently.

From Pipelines to AI Pipelines – Smarter Flexibility for Data Analytics

This is where AI Pipelines come into play. They extend the existing pipeline concept with AI-assisted code generation, merging the intuitive workflow of the Pipeline Editor with the flexibility of open programming.

The process is simple and efficient:

1. Select a data stream Choose the relevant sensor or process data – for example, temperature, pressure, or vibration signals.

2. Choose a programming language Currently, Python and JavaScript are supported – two widely used languages with extensive ecosystems.

3. Describe your goal Using a natural language prompt, users describe what they want to achieve – e.g. “Detect anomalies in the pressure curve and trigger an alert if the mean value deviates by more than 10% from the reference.”

4. Automatic code generation A Large Language Model (LLM) automatically generates the corresponding code and integrates it directly into the pipeline.

5. Explain and refine On request, the generated code can be explained in natural language – a great help even for users with basic coding knowledge.

6. Run the pipeline live Finally, the AI Pipeline is executed in real time on the incoming machine data.

Within minutes, users can create fully customized analyses – without deep technical expertise, lengthy configurations, or external development effort.

Benefits

AI Pipelines offer a new level of flexibility, transparency, and ease of use:

• More freedom: Create custom logic and algorithms directly in code — no limits from predefined blocks.
• Simpler interaction: Describe what you need in plain language instead of configuring complex settings.
• Full transparency: The generated code is visible, understandable, and editable at any time.
• Faster results: Move from idea to running live analysis within minutes.
• Learning support: Auto-generated explanations help users understand the logic and improve coding skills.

Try it now!

The new AI pipelines make the Bytefabrik IIoT Application Platform even more powerful—and easier to use at the same time. They enable specialist users to create their own monitoring rules, condition monitoring analyses, and data processing logic—supported by state-of-the-art AI technology.

This sets a new standard in industrial data analysis: maximum flexibility combined with minimum complexity.

The AI pipelines are available now and are part of our product Bytefabrik IIoT Application Platform.

If you are interested in a live demo or individual consultation, please contact us – we will show you how AI can take your data analysis to the next level.