What is the first step you take to set up a successful AI initiative in a new organization or domain, and why?

Most answers emphasize defining a clear problem or objective as the first step, while others focus on leadership buy-in, process readiness, data quality, or building foundational AI understanding.

12 answers

Answers

  • Michael Bruun

    Partner, Global co-Head Private Equity, Goldman Sachs Asset Management · Goldman Sachs

    Get CEO buy-in

    Michael Bruun says it needs to start with CEO buy-in, because it has to come from the top: the CEO must make clear to all employees that AI is an absolute must.

    The answer starts at 32:30.

    How I Invest with David Weisburd · Listen on Apple Podcasts
  • Amit Mathradas

    Chief Executive Officer and Board Member · Five9

    Identify the problem

    Amit Mathradas says the first step is to figure out what problem you are trying to solve, since many people rush to throw AI at a problem before considering whether AI actually solves it.

    SiliconANGLE theCUBE · Watch on YouTube
  • Vinod Muthukrishnan

    VP / GM Webex Customer Experience · Cisco

    AI readiness assessment

    Vinod Muthukrishnan says the first job is to help people put points on the board, pointing to a document on AI readiness assessment, so teams can learn empirically at a small scale before scaling up.

    SiliconANGLE theCUBE · Watch on YouTube
  • Autumn D. Krauss, Ph.D.

    Chief Scientist, Future of Work Research Lab · SAP SuccessFactors

    Build AI literacy

    Autumn D. Krauss, Ph.D. says the first step is building AI literacy, which she describes as a foundational understanding of how AI works, including its ethics and fit-for-purpose use cases, rather than training on specific tools.

    NZ Tech Podcast · Watch on YouTube
  • Conny Ploth

    AI Transformation Vice President, Global Data & AI · Santander

    Fix the process

    Conny Ploth says the first step is to fix the process, because putting an AI layer on a messy process only speeds up inefficiencies instead of resolving their root cause.

    The answer starts at 4:45.

    Marketer of the Month · Listen on Apple Podcasts
  • Paula Goldman

    Chief Ethical and Humane Use Officer & EVP, Product · Salesforce

    Pick a few big bets

    Paula Goldman says the first step is to pick a few big bets and go all in on them, because most companies already have enough data for that.

    The answer starts at 27:48.

    Thinkers & Ideas · Listen on Apple Podcasts
  • Sergio Torres

    Senior Vice President Product Management · Descartes Systems Group

    Define the first decision

    Sergio Torres says to start by defining the first decision the agent should make, because that decision dictates what data is non-negotiable and other things are distractions until that decision works.

    SupplyChainBrain · Watch on YouTube
  • Neil Morelli, PhD

    Senior Director, Human-AI Collaboration and Workforce Transformation · Salesforce

    Start small with a business case

    Neil Morelli advises starting small but with a higher-level business case that keeps the big picture and a strong why in mind, because that builds more momentum.

    The Science of Excellence · The Science of Excellence · Listen at 28:59
  • R "Ray" Wang

    Principal Analyst, Founder & Chairman · Constellation Research, Inc.

    Define the problem

    R "Ray" Wang says to define the problem and understand its stakeholders, so the cool solution does not take over.

    DisrupTV · DisrupTV · Listen at 14:21
  • Andrew Embry

    Sr. Director Insights Innovation Capabilities- Global Core Market Research · Eli Lilly And Company

    Identify a real problem

    Andrew Embry says the first part of AI adoption that sticks is to identify a real, tangible problem for AI to solve, since claiming AI can do everything means it can’t really do anything.

    The Curiosity Current: A Market Research Podcast · The Curiosity Current: A Market Research Podcast · Listen at 10:18
  • Valentina Jordan

    CEO & Co-Founder · Nauta

    Define a clear objective

    Valentina Jordan says the first step is to have a very clear objective, because without a clear expectation of the work it will never give ROI.

    TalkingLogistics · Watch on YouTube
  • JOSE MARTINEZ

    Head of Costing Engineering and Program Control · Nissan Do Brasil

    Ensure clean, curated data

    Jose Martinez says you need clean, curated data that works and has the right value to run statistical tests, machine learning or LLM models; with unnormalized or duplicated data, you won't get the result you're looking for.

    The Agile Brand with Greg Kihlström®: Expert Mode Marketing Technology, AI, & CX · The Agile Brand with Greg Kihlström®: Expert Mode Marketing Technology, AI, & CX · Listen at 16:21