Christine Livingston, Perficient’s own Chief Strategist for AI, presented at a session on real-world success stories in AI at AI Summit. As always, she focused on the best practices in how this works.
Quote: AI adoption has tripled in the past 3 years. We’ve barely scratched the surface
What is AI really?
- Machine Learning
- Natural language processing
- Predictive Analytics
- Cognitive computing
- Signal Services
Note: RPA is typically a solution that works hand in hand with AI to achieve automation. it’s not the intelligence behind it.
Key Reasons AI programs fail
- There is no strategy. It’s undefined and ad-hoc
- Business Value is misunderstood
- Culture: unwilling to adapt and unprepared for rapid transformation
- Scale: only deployed for a single-use case. Not aligned with teams for more change
Developing a strategy
Think Big, Start Small, Act Fast
You work from Current state to business alignment to technology alignment and then create a candidate roadmap. it’s in that order. Note the business needs drive your strategy. As part of this, you should prioritize your use cases.
With Technology Alignment you will create a reference architecture.
Creating the roadmap:
- Evaluate and prioritize
- Decompose it. Create your list. Translate to a foundational capability. This will drive your plan
- Execute. Don’t forget to start small
- prove, deploy, and iterate
So who owns this?
You should create a center of excellence. If you want to scale, then you need to create a hub of expertise.
Individual Bank: Use the strategic approach to create the prioritized roadmap. Started with a document processing framework for intelligent automation. brought a lot of solutions together like Kofax and Alfresco. They automated the extraction with natural language processing and visual recognition.
Wanted to change the customer experience. It’s hard to find the right car online. Compare that to walking to a dealer and giving them your needs. They wanted to merge the physical and digital experience. They created an intelligent assistant to guide consumers through the process to get real-time product recommendations. This was always on btw. The conversational AI never sleeps.
Note: this same AI converted to an Alexa skill as well
How does this relate to Financial Services: You have a lot of products with very odd names that no non-professional understands. This is a perfect use case for what Hyundai did.
Wanted to know what factors drive a patient to being readmitted. Using traditional data fields, they could only get to a 70% success rate. Getting above that meant getting into the unstructured data in notes. Doing this drove a much higher ability to predict who needed more support
Used an intersection of search and AI. Symantec redid their knowledge base and integrated a virtual agent.
Want to learn more about the latest discussions at AI Summit? Read more about what is happening at AI Summit here.