PhenoWorks beta is here: a crop phenotyping web platform that brings research data, reusable analysis methods, and an AI assistant into one workspace. Load and visualize your data, extract structured features with LgoPy blocks, and use the results to support analysis, interpretation, and research writing.

Our goal is to make research data analysis easier to repeat, inspect, and share—from the first exploratory run to a colleague revisiting the experiment months later.

Last month, we outlined a roadmap for bringing trained models into LgoPy workflows. That design has now reached its first working implementation in PhenoWorks.

A researcher can register a packaged inference service, validate it, deploy it in an isolated container, and use it from an LgoPy block through an authenticated gateway. The block refers to a cataloged model version; it never needs to carry an endpoint, credential, or container detail in the analysis code.

Crop phenotyping research rarely stays in one application. You might organize field trials in PhenoWorks, explore measurements in a Python notebook, and fit a statistical model in R. Moving between those tools should not mean repeatedly downloading files by hand or writing the same HTTP requests for every project.

The PhenoWorks SDKs for Python and R bring projects, datasets, files, and analysis pipelines into your existing scripts. Browse data, submit a workflow, wait for its results, and download the outputs for your next analysis—all through your PhenoWorks account.

Models are an essential component of modern data science workflows, from traditional machine learning models built on engineered features to deep learning models and large language models (LLMs). Yet a model can work well in a research notebook and still be difficult for a colleague to reuse. The trained weights are only part of the method. Image normalization, feature order, class definitions, library versions, and the code that turns predictions into usable measurements all affect the result.

The current implementation already allows a custom LgoPy block to use a model by calling its API endpoint. The block author supplies the connection logic and adapts the model's inputs and outputs to the workflow. We believe PhenoWorks can build on this capability by providing a standard way to package and reuse models within LgoPy pipelines, just as data analysis algorithms can already be packaged as reusable LgoPy blocks.

An image-processing script works well on your first dataset. Months later, a new field campaign brings another batch of images, and you need to repeat the analysis. Which version of the script did you use? What settings produced those outputs? Could a colleague reproduce the same steps from the files you shared?

Reproducible data processing depends on preserving those details alongside the method. LgoPy blocks help by packaging a transformation with declared inputs, configurable parameters, and versioned code. In this tutorial, we will use a simple RGB-to-HSV conversion to show how data reaches a block from each upload method, how a block declares the sensors and data products it reads, and how to save derived outputs that can be filtered and reused across datasets in PhenoWorks.