Implementer’s Guide to AI

Increasing revenue where AI unveils new market opportunities and streamlines operationsReducing costs where, through automation and process optimisation, AI reduces operational costs and enhances overall business efficiencyDriving customer loyalty where AI creates personalised experiences to help customers feel valued and understood, which builds and maintains loyaltyReducing risk or detecting fraud to enhance business resilience

	Exploring	Experimenting	Implementing	Deployed	Not Applicable
AI-generated emails/content drafts to improve the productivity of sales, service and marketing teams
Using AI recommendations to personalise offers for customers
Automating service leveraging AI-powered chatbots / virtual assistants
AI-based auto summarisation for documents, emails, conversations, etc.
Auto-detect fraud by analysing data, behaviour and patterns

Yes - We have integrated data from all internal and external sources to create a single view.No - Customer data (marketing touchpoints, sales interactions, service history, purchase history) resides in different repositories.Partially - Some systems are integrated, while others are not.

Responsible AI practices not in placeBasic awareness without a formal, responsible AI frameworkDeveloping awareness and formal steps towards responsible AI-initiatedMatured responsible AI practices and policies

	Most Confident	Moderately Confident	Least Confident
Data Protection – Using data safeguards like data minimisation, on-device learning, federated learning, etc.
Diverse Development – Creating models with diverse teams and high-quality, representative datasets
Monitoring – Monitoring models before and after deployment with human supervision
Cybersecurity – Implementing security measures against unauthorised access and adversarial attacks
Documentation – Maintaining documentation about system capabilities and data sources for scrutiny
Explanation – Providing easy-to-understand explanations for sensitive or significant model decisions
Liability – Addressing liability for AI-related harms fairly with grievance mechanisms
Human-Centered Design – Focusing on aligning model development and use with human value
Compliance – Adhering to applicable laws and regulations

YesNoNot sure about its utility

	Most Challenging	Moderately Challenging	Least Challenging
Lack of high-quality data
Lack of skilled technical and management personnel
Regulatory uncertainty (i.e., lack of clarity on applicable laws, rules, and regulations)
AI vendors who are not committed to ensuring compliance with responsible AI standards
High cost of implementation

	Most Concern	Moderate Concern	Least Concern
Data Privacy - Sending company data on public LLMs
Hallucinations and reliability of AI responses
Data Bias and Toxicity

By submitting this form, you agree to have your contact information, including email, passed on to Salesforce for the purpose of following up on your interests and you that you have read and agree to Research NXT's Privacy Policy, and also the Salesforce Privacy Statement.