Deploying and managing AI assets

Use IBM Watson Machine Learning to deploy and manage AI assets, including models and generative AI assets, and put them into pre-production and production environments. You can monitor the deployed assets for fairness and explainability. Further, you can orchestrate the end-to-end flow of assets from creation through deployment by using IBM Watson Pipelines.

Deploying AI assets and orchestrating pipelines

Deploying your model to the production environment is a crucial step that you must follow after building or importing your model.

AI lifecycle

Deploy model

You can use Watson Machine Learning to deploy, monitor, and manage your AI assets by promoting your assets to a deployment space. You can also use watsonx.ai to deploy tuned foundation models and prompt templates. After deploying your model, you can evaluate your model within your deployment space for fairness, quality, and drift.

For more information, see Deploying AI assets.

Automate pipeline

You can automate the path to production by building a pipeline to automate parts of the AI lifecycle from building the model to deployment, and evaluate a model to shorten the time from conception to production by using Watson Pipelines.

For more information, see Orchestrating tasks with Pipelines.

Managing AI lifecycle with ModelOps

You can organize and manage assets through the development, testing, and production phase of the AI lifecycle. Further, you can share data assets in a feature store, automate ModelOps by using Pipelines, track models with AI factsheets, and evaluate model deployments.

For more information, see Managing the AI Lifecycle with ModelOps.

Tutorials and use cases

The following resources demonstrate how to plan for managing machine learning assets and how to build key pieces of your solutions.

Review quick start tutorials to learn how to build and deploy machine learning models:

Quick start tutorials
Tutorial	Description
Build and deploy a machine learning model	Build a machine learning model by using a no-code approach with an automated tool, AutoAI. Deploy the trained model by creating a batch deployment. Score your batch deployment by creating a batch job.
Build an deploy a machine learning model in Jupyter Notebook	Build a machine learning model by using code in a Jupyter notebook to prepare data, build your model, and deploy your model.
Build and deploy a machine learning model with SPSS Modeler	Build a machine learning model by using a no-code approach by using an SPSS Modeler flow. Deploy the trained model by creating an online deployment and test the model in real-time.
Build and deploy a Decision Optimization model	Build a machine learning model by using Decision Optimization.
Evaluate a machine learning model	Deploy a model, configure monitors for the deployed model, and evaluate the model.

AI governance

The following tutorials use the scenario for Golden Bank, that needs a model to identify whether customers qualify for mortgages to reduce the bank’s application processing costs.

AI governance use cases
Use case	Description
Build and deploy a model	Train a model, prompt it to a deployment space, and deploy the model.
Test and validate the model	Evaluate a model for quality, fairness, drift, and explainability.

Data Science and MLOps

The following tutorials use the scenario for Golden Bank, that needs to automate a data pipeline that delivers up-to-date data on all mortgage applicants, that lenders can use to make decisions.

Data Science and MLOps use cases
Use case	Description
Orchestrating an AI pipeline with model monitoring	Create an end-to-end pipeline that transforms data and trains a model.
Orchestrating an AI pipeline with data integration	Create an end-to-end pipeline to prepare your data, train your model, and validate your model.

Learn more

Parent topic: Analyzing data and working with models

Deploying and managing AI assets

Deploying AI assets and orchestrating pipelines

Tutorials and use cases

Related quick start tutorials

Related use cases

Learn more