Product Management for AI

Book description

The increasing push to develop products that integrate AI puts the intersection of AI and product management into sharp focus. AI brings many challenges to traditional product management, including nondeterministic outcomes and the potential for bias against particular groups. These problems aren't insurmountable, but they're real, and they cause many projects to fail before they're deployed.

In this report, authors Justin Norman, Pete Skomoroch, and Mike Loukides present four in-depth essays to help business leaders, AI specialists, and data scientists examine what makes AI different. Once you're familiar with the issues, you'll be better prepared to anticipate and solve the problems you face as you develop an AI project and shepherd it into production.

Originally published in O'Reilly Radar, each of these essays provides helpful supporting examples. Essays include:

  • "What You Need to Know about Product Management for AI," by Pete Skomoroch and Mike Loukides
  • "Practical Skills for the AI Product Manager," by Pete Skomoroch, Mike Loukides, and Justin Norman
  • "Bringing an AI Product to Market," by Pete Skomoroch, Mike Loukides, and Justin Norman
  • "AI Product Management after Deployment," Mike Loukides and Justin Norman

Table of contents

  1. Preface
    1. Acknowledgments
  2. 1. What You Need to Know About Product Management for AI
    1. Why AI Software Development Is Different
    2. Machine Learning Adds Uncertainty
    3. AI Product Estimation Strategies
    4. Organizational Prerequisites for AI at Scale
    5. Identifying Viable Machine Learning Problems
    6. Work on Things That Matter to Your Business
    7. Prioritizing with the Business in Mind
  3. 2. Practical Skills for the AI Product Manager
    1. The AI Product Pipeline
      1. Innovation/Ideation/Design for UI/UX
      2. Feature Development and Data Management
      3. Experimentation
      4. Research
      5. Modeling
      6. Serving Infrastructure
    2. Consumer Companies Versus B2B Companies
    3. Startups Versus Large Companies
    4. The Data Expertise of the AI PM
      1. Skill–Data Lifecycle and Pipeline Management
      2. Skill–Experimentation and Measurement
      3. Skill–DS/ML/AI Development Process
    5. Conclusion
  4. 3. Bringing an AI Product to Market
    1. The Core Responsibilities of the AI Product Manager
      1. Identifying the Problem
      2. Addressing the Problem
      3. Fault-Tolerant Versus Fault-Intolerant AI Problems
      4. Planning and Managing the Project
    2. The AI Product Development Process
      1. Understand the Customer and Objectives
      2. Data Exploration and Experimentation
      3. Data Wrangling and Feature Engineering
      4. Modeling and Evaluation
      5. Deployment
      6. Monitoring
    3. Executing on an AI Product Roadmap
      1. AI Product Interface Design
      2. Picking the Right Scope
      3. Prototypes and Data Product MVPs
      4. Data Quality and Standardization
      5. Augmenting AI Product Management with Technical Leadership
    4. Testing ML/AI Products
    5. Conclusion
  5. 4. AI Product Management After Deployment
    1. Debugging AI Products
      1. I/O Validation
      2. Inference Task Speed and SLOs
      3. Durability
      4. Monitoring
    2. Post-Deployment Frameworks
    3. Where Do We Go from Here?
  6. A. Additional Resources

Product information

  • Title: Product Management for AI
  • Author(s): Justin Norman, Peter Skomoroch, Mike Loukides
  • Release date: February 2021
  • Publisher(s): O'Reilly Media, Inc.
  • ISBN: 9781098104191