Use Cases & AI Entry Points¶
Overview¶
Organizations adopt AI on IBM Z differently based on their business goals. Some may adopt a pre-built solution, deploy an existing model, or build their own. Understanding these entry points can help you identify the products, skills, and lifecycle stages most relevant to your AI journey.
The Three AI Entry Points¶
While AI implementations vary, many follow a similar path from identifying a business problem to delivering AI-driven outcomes. The three entry points below reflect common approaches organizations take when adopting AI on IBM Z.
-
1 — Turnkey Solution
Install, configure, deploy.
IBM and its partners offer pre-built AI solutions that are ready to deploy with minimal setup. These solutions embed AI directly into the platform or provide vendor-managed models that are purpose-built for specific business domains.
When to choose this entry point
- Your team has limited or no data science expertise
- You need fast time-to-value with minimal build effort
- A proven, pre-built solution already exists for your use case
- You prefer a managed or embedded AI experience
-
2 — Select a Model via IBM or Vendor
Leverage an existing model with IBM Z inferencing acceleration.
Choose a foundation model, open-source model, or vendor-supplied model and deploy it on IBM Z using IBM's inferencing infrastructure and hardware acceleration. No model training required — focus is on serving and integration.
When to choose this entry point
- A suitable model already exists (IBM, open source, or vendor)
- Your team has integration and deployment skills but not model training expertise
- You want hardware-accelerated inferencing (Telum, Spyre) without building a model from scratch
- You are deploying a foundation model or LLM for generative AI workloads
-
3 — Build Your Own Model
Full AI lifecycle — from your data to your deployed model.
When no existing model meets your needs you can build one. This entry point covers the full lifecycle: find and prepare data, train a model on your enterprise data, test and validate it, optimize it for deployment, and serve it in production on IBM Z.
When to choose this entry point
- Your use case is highly specific to your business and data
- You have data science and machine learning expertise in-house
- Existing models do not provide sufficient accuracy or coverage
- You need full control over model behavior, training data, and versioning
Which Entry Point Is Right for You?¶
| If you want to… | Entry Point |
|---|---|
| Adopt AI quickly with no model training or data science team | 1 — Turnkey Solution |
| Deploy a proven IBM, open-source, or vendor model on IBM Z hardware acceleration | 2 — Select a Model |
| Train a model on your own data and own the full AI lifecycle | 3 — Build Your Own |
Once you've identified your entry point, explore the AI Lifecycle for implementation guidance, and Products & Tools for a full catalog of available technologies.
Additional Use Case Resources¶
-
Advanced Credit Card Fraud Detection
See a full end-to-end solution blueprint — Advanced Credit Card Fraud Detection using an XGBoost and BERT ensemble on IBM z17.
-
Explore more AI solution patterns covering generative AI, fraud detection, credit risk, health insurance claims, data preprocessing, and anti-money laundering.
-
Follow the end-to-end AI lifecycle on IBM Z — from preparing data and training models to deploying, serving, and governing AI in production.
-
Browse the full catalog of IBM Z AI products and tools across the lifecycle to identify what fits your use case and entry point.