Key concepts in Decision Intelligence

DIP (Decision Intelligence Platform):

A platform that encompasses all elements of decision-making, including business goals, decision-making processes, business decisions and actions, data, AI, human experts, rule systems, and applications. Well-informed business decisions are made through the orchestration of all these elements. The need for DIP arises from drivers, such as increased business uncertainty, siloed decisions, fragmentation of decision-making components, collaboration between AI and humans, regulations, and more.

We are addressing key questions to guide the implementation of DIPs.

1) Business KPIs   

- What are the business KPIs?   

- How are the KPIs related to each other?   

- How are the KPIs structurized hierarchically?

2) Business Decision Processes   

- Who are the participants in the decision-making process?   

- What decision-making methods are currently used?   

- What data is related to the decision-making process?

3) Business Decisions   

- What are the business decisions/actions?   

- What are the unit environments where decisions take place?   

- What are the business constraints?

Business KPI: 

A shared, measurable goal that all decision-making participants and elements aim to achieve together. The concept of a KPI is universally well-known, yet often overlooked. AI is typically evaluated by statistical criteria, which may not be directly related to business and cannot be explainable for business decisions. Without a measurable goal, neither human participants nor AI and autonomous systems can effectively know where to aim. As decisions across a company impacting one another, it is crucial to design and clarify the relationships among various Business KPIs to implement decision intelligence. Business KPIs help participants understand, improve, and monitor business decisions.

Decision Model:

A unified function that produces decisions & estimated KPIs. This model can integrate all decision-making elements such as business KPIs, business rules, AI models, and human experts. Current AI techniques typically conclude at the narrow-contented modeling step and often fail to integrate well with business decision-making process. Additionally, other critical factors like business goals, constrains, competition, regulations, and decision are not adequately addressed. Fundamentally AI outputs are not inherently suitable for the requirements of business decisions.To be effective, a decision model must meet five essential requirements: 1) Aligning all required decision-making elements 2) Directly supporting complex business decisions 3) Explainability in terms of business decisions 4) Adjustability based on business goals and considerations 5) Adaptability to the business environment in a timely manner

Decision Governance:

Decision Governance is a tool that enables a company to build a comprehensive, company-wide decision-making structure and respond to changes promptly.

Business decision-making is a complex and dynamic process. It affects not only business outcomes but also involves risks related to confidentiality, regulations, privacy, and more. Additionally, decisions across a company are interconnected and impact each other, necessitating timely adjustments. Business decisions change as frequently as changes the business environment.

Governing decisions requires three key functionalities: structuring, monitoring and updating decisions.

1) Structuring: to govern decisions, structuring business decision participants’ authorizations includes roles for both humans and AI. This covers not only decision modeling also approval process, regulations and audits.

2) Monitoring: it encompasses the entire business operation. It involves monitoring business status through KPIs, issuing warnings and suggestions, tracking the concordance rate between primary decisions and human’s decisions, and maintaining model status.

3) Updating: Often confused with regular model retraining, updating in business operation involves adjusting and updating decisions on a daily, weekly or monthly basis. Therefore, updating the decision model must be completed by business units in a timely manner.