Cassie Kozyrkov

Why organisations work with Cassie Kozyrkov

• She built the Decision Intelligence function inside Google from scratch and ran it for five years, which means leaders get advice from someone who has already solved, at scale, the problem they are trying to solve.
• She named and codified the field itself. When a CEO asks what Decision Intelligence is and why their AI investment should sit inside it, she is the primary source, not an interpreter of someone else’s framework.
• She personally trained more than 20,000 Googlers across every function in data-driven decision-making and AI, so she knows how to make the material land with non-technical executives as well as engineers.
• She translates AI and statistics into decision-language that boards and operators can use. Her Making Friends with Machine Learning course, originally built only for Google, is now the reference explainer for thousands of leaders outside the company.
• She advises at the level the material requires. Her Federal Reserve Bank of New York Innovation Advisory Council seat and her work with Forbes on AI signal that her analysis is taken seriously by policy and media institutions, not only by corporate buyers.

Biography highlights

• Google’s first Chief Decision Scientist (2018 to 2023), leading the firm’s decision intelligence practice inside Research and Machine Intelligence.
• Founder and CEO of Kozyr, an AI advisory and education business serving senior leaders and their teams.
• Personally trained more than 20,000 Googlers in data-driven decision-making and AI, and guided over 500 decision intelligence projects inside Google.
• Creator of Making Friends with Machine Learning, originally an internal Google course, now a free public course on YouTube.
• Author of the Decision Intelligence Substack and LinkedIn Top Voice for six consecutive years; writing has appeared in Harvard Business Review, Fortune, Fast Company, WIRED, and The Wall Street Journal.
• Member of the Federal Reserve Bank of New York’s Innovation Advisory Council (2023 to 2024); Duke University Few-Glasson Distinguished Alumna.

Biography

Most companies do not have a data problem, they have a decision problem. The dashboards, pipelines, and models are in place, and yet senior leaders still cannot say which decisions are demonstrably better because of any of it. That gap is the subject Cassie Kozyrkov has been working on for a decade, first inside Google and now with leaders across industries.

At Google, she became the company’s first Chief Decision Scientist in 2018 and held the role until 2023. She built a new function that sat between Research and Machine Intelligence on one side and the operating business on the other, and used it to run AI and data through a discipline she calls Decision Intelligence. By the time she left, she had personally trained more than 20,000 Googlers and influenced over 500 projects against that framework.

The reason her work travels beyond Google is that she made the material legible. Her course Making Friends with Machine Learning, built inside Google and later released in full on YouTube, is one of the few genuinely plain-English explainers of how machine learning actually fits into business decisions. Her writing in Harvard Business Review, Fortune, Fast Company, WIRED, and The Wall Street Journal has done the same for non-technical executives, and her Decision Intelligence Substack has become a working reference for practitioners.

Today she runs Kozyr, advising leadership teams on AI strategy and decision design, and sits on the Federal Reserve Bank of New York’s Innovation Advisory Council on AI and financial markets. The through-line is consistent. She is the rare AI voice who starts with the decision the organisation is trying to get right, and works backwards to the technology, rather than the other way round.

Key speaking topics

• Decision Intelligence
• AI strategy and adoption
• Data-driven decision-making
• Machine learning for business leaders
• Responsible and effective use of AI
• AI literacy for executives
• Bias and judgement in AI systems

Ideal for

• CEOs, boards, and executive teams setting AI strategy and capital allocation against it
• Chief Data, Analytics, and AI Officers building a decision-intelligence operating model
• Transformation and strategy leaders responsible for turning AI pilots into business results
• Leadership programmes and offsites that need genuine AI literacy, not a vendor pitch

Audience outcomes

• A working definition of Decision Intelligence and where it sits relative to data science, analytics, and AI investment.
• A sharper diagnostic for why AI projects stall: wrong decision scoped, wrong owner, or wrong question asked of the model.
• Shared language across technical and non-technical leaders for debating AI initiatives without defaulting to buzzwords.
• Specific techniques for pressure-testing decisions against common biases, including outcome bias and confirmation bias.
• A realistic read on what current AI systems can and cannot be trusted to decide without human judgement in the loop.