Brief Summary
This workshop, the second in a three-part series, explores business strategies for AI adoption, focusing on how AI can shape and enhance business strategies. It covers the role of AI in brainstorming, data-driven insights, and identifying business opportunities. The session also differentiates between traditional and big data analytics, highlighting the importance of predictive analytics for smarter decision-making, and examines AI-driven market analysis and competitive intelligence. Case studies from companies like Klarna, BMW, and Moderna illustrate the practical benefits of AI integration.
- AI facilitates quick brainstorming and combines ideas from different domains.
- Big data analytics is data-driven, uncovering correlations and actionable insights.
- High-quality data is crucial for accurate AI model predictions.
Introduction
The workshop is the second in a series focusing on AI in business, specifically strategies for AI adoption. Mia Kennady, co-founder and CTO of Venture Factory AI, is introduced as the speaker. The session aims to explore how AI can help formulate business strategies, covering topics from AI's role in brainstorming to using AI for market analysis and competitive intelligence.
AI's Role in Business Strategy
AI, particularly language models, can quickly brainstorm business ideas by drawing on collective web experience. While AI excels at combining ideas from different domains, it struggles with creating entirely novel concepts. AI provides data-driven insights, uncovers profit drivers, analyses effectiveness, and monitors businesses in real-time. To spot AI opportunities, focus on outcomes that significantly impact the business, relevant data points, and the risk associated with incorrect predictions, starting with lower-risk initiatives.
AI Impact by Industry
The impact of AI varies by industry, with retail leading due to extensive e-commerce data. Transport and logistics benefit from AI in supply chain management and logistics improvement. Automotive and assembly see gains from predictive maintenance and robotic process automation. Banking and high-tech industries also experience significant AI impact due to their existing tech infrastructure and data availability.
Traditional Analytics vs. Big Data Analytics
Big data exceeds the capacity of conventional methods, requiring significant storage and tools for visualisation and analysis. Traditional analytics starts with a question and hypothesis, collecting data to clarify it. Big data analytics, however, is data-driven, exploring data to uncover correlations and actionable insights. This approach allows for asking new questions and answering existing ones more effectively. Preparing data is a significant part of the workload in big data analytics, including acquisition, verification, and formatting for AI models.
Life Cycle of an ML Project
The life cycle of a machine learning (ML) project involves planning, data collection and preparation, model training and debugging, and deployment with testing. This process is iterative, requiring adjustments based on insights gained at each stage. It's crucial to factor in the iterative nature of ML projects when planning and budgeting time and costs.
Predictive Analytics for Smarter Decision Making
AI enables smarter decisions through data-driven planning and better risk management. While AI can forecast based on historical data, understanding causation is essential for making the right decisions. Experimentation, such as A/B testing and multi-arm bandit algorithms, helps compare action versus no action to understand the true impact of decisions.
What Drives Prediction Accuracy
Prediction accuracy in AI models is driven by data quality and quantity, model choice, and domain expertise. High-quality data is particularly impactful. When improving model performance, investing in better data often yields greater results than experimenting with different models.
AI-Driven Market Analysis and Competitive Intelligence
AI enhances competitive intelligence by monitoring competitor pricing and product offerings at scale. It analyses market data, including financial data and customer feedback, to benchmark competitive performance. For market analysis, AI mines data from social media, news, and e-commerce platforms to understand customer preferences and upcoming trends. AI also helps analyse product market fit by assessing execution capabilities, potential products, and market needs.
Case Study: Klarna
Klarna, a financial services company, integrated generative AI early on, using an AI assistant named Kiki to handle a significant portion of customer service calls in multiple languages. Klarna connected data across the organisation, removing siloed data architecture to link people, projects, documents, teams and processes to outcomes. Despite initial success, Klarna learned that some customer service calls still require human agents, highlighting the importance of easy access to human support.
Case Study: BMW
BMW uses AI for predictive analysis of global sales, consumer behaviour, and inventory optimisation. They empower their workforce with AI assistants and tools to build their own apps, boosting productivity and innovation. BMW also has strict data life cycle management, linking use cases to data assets to measure their value. Early wins included data centralisation and strong data governance, leading to improved product market fit, enhanced customer satisfaction, and increased operational efficiencies.
Case Study: Moderna
Moderna integrated AI and analytics early on, with a top-down strategic intent communicated across the company. They invested in employee upskilling through an AI academy and empowered employees to build custom GPTs. Strategic partnerships with OpenAI and AWS have been crucial, with employees using custom-built AI chatbots and AWS's OC3 for customer service.
Q&A: Creating Chatbots and Privacy Concerns
The Q&A session covers creating chatbots by training them with existing conversations and fine-tuning them with domain-specific terms. It also addresses privacy concerns, noting that data sent to public language models is used for further training. Organisations dealing with sensitive data may opt for locally hosted language models.
