Getting ready for an AI project at your business? Cool, what kind? An essential step towards AI success is to understand which usage pattern will address your business need. There are seven key patterns, summarized below.
Each of the project types below use artificial intelligence in some way. However, each also has different key concerns necessary for success. Knowing and navigating the kind of project you have is important!
| Project Type | Examples | Timeline | Important |
|---|---|---|---|
| Autonomous Systems | Self-driving cars, Robots serving hamburgers. Also, a steering wheel that avoids crossing a highway line and an agent that can perform administrative tasks for you | Longest time to ROI. Still mostly an emerging tech | Long term horizon, a clear understanding of the machine intelligence level needed |
| Conversation & Human Interaction | Chatbots for Customer Services and reference assistance (coding, paralegal) | Shortest time to ROI, most commidified pattern | A clear-cut Q&A-type business process to augment. Particularly, customer support areas or others driven by a knowledgebase |
| Goal-driven systems | Game playing systems, an agent that bids at auctions | Long time to ROI, little commodification to jumpstart your build | Clear understanding of the goals and operating space |
| Hyper-personalization | User-specific storefronts, educational content, news... | Medium time to ROI | Customer focus |
| Patterns & Anomalies | Fraud detection, maintenance early warnings | Medium time to ROI | Being able to map your own needs conceptually to a foundation model, to speed delivery; attention to validation during model training |
| Predictive Analytics & Decisions | Predicting maintenance needs, stocking levels, watching markets and competition | Medium time to ROI | (Even more than usual) Attention to validation during model training, long-term plan for watching "data drift" that can skew results over time |
| Recognition | Labeling/detecting objects in images, Handwriting to text | Can be short, lots of commodification ("foundational models") | Ability to map own need conceptually to a foundation model, to speed delivery |
How do you know which type of project you have? The first step to any AI-enabled project's success is to clearly define your business goals. Yes, AI projects are technology projects. And, a key success factor is understanding the need that technology solves.