[DSC DACH 24] AI, how to go smart about it - Milos Solujic

In today's rapidly evolving business landscape, Artificial Intelligence (AI) stands out as a transformative force. However, navigating the AI terrain can be daunting, particularly for medium and large business executives. This talk offers a pragmatic approach to integrating AI into your organization, highlighting common pitfalls and providing a step-by-step guide for smarter adoption. Drawing from our extensive experience in AI implementation across various industries, we'll uncover typical mistakes that companies often make—such as overestimating capabilities, underestimating costs, neglecting data quality, and failing to align AI initiatives with business goals. These errors can lead to wasted resources and missed opportunities, but they are entirely avoidable with the right strategy. Our session will guide you through a smarter, step-by-step approach to AI adoption: Assess Readiness: Understand your current capabilities and set realistic expectations, Define Clear Objectives: Align AI initiatives with strategic business goals, Prioritize Data Quality: Ensure your data is accurate, relevant, and secure, Start Small: Begin with pilot projects to validate concepts and demonstrate value, Foster a Culture of Innovation: Encourage cross-functional collaboration and continuous learning, Monitor and Iterate: Regularly review progress, adapt strategies, and scale successful initiatives.

1. AI How to go smart about it This talk offers a pragmatic approach to integrating AI into your organization, highlighting common pitfalls and providing a step-by-step guide for smarter adoption. Milos Solujic CIO Opcom.io Sep 13th 2023

2. One Strategy: Fake it

3. Who am I? 20 years in software and data science Most of it running consulting shop that helps businesses with modern challenges through technology opcom.io msolujic

4. Agenda What AI? Some paths toward failure with AI A Framework for success, examples Future Recap

5. What is AI ? Artificial intelligence (AI) is technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making , creativity and autonomy .

6. What kinds of AI are there?

7. What kinds of AI are there?

8. LLMs and its ancestors

9. Some pitfalls with Data and AI Initiatives Lack Of In-House Expertise Uncertainty About Where To Implement It The Absence Of Capable Infrastructure Lack of Reliable and Structured Data Data Privacy And Security Concerns Technological Overwhelm The Sheer Number Of Options Available

10. ‹#› AI Strategy? How about Data Strategy …

11. What are main issues with data Data Silos Diverse tech used Expensive to scale Expensive to operate Data quality Governance

12. Trap 1 Correlation IS NOT a Causation

13. Trap 2 Collect than what?

14. Trap 3 Fall in love in your idea and than look for data/AI that will confirm it

15. ‹#›

16. FOMO is not worst…

17. Once you grast AI concepts, and where are the traps, you could follow this: Explore and Pick a business problem Assess Readiness Define Clear Objectives Monitor and Iterate Scale up AI initiatives, share knowledge A Framework for success

18. Pick a solid business problem that will de-risk further AI adoption

19. Look around for the suitable business function

20. Pick SMART: For example, low hanging fruit of business problems that can be tackled via data or AI Data Availability and quality Business impact Feasibility Risk assessment Pick first problem for POC

21. Example Area = Sales, Example Process = RFP Management

22. Visualize problem area Business impact Feasibility

23. Example Area = Sales, Example Process = RFP Management

24. Pick SMART: For example, low hanging fruit of business problems that can be tackled via data or AI Data Availability and quality Business impact Feasibility Risk assessment Pick appropriate AI technique Begin with POC/pilot projects to validate concepts and demonstrate value. Pick first problem for POC

25. Assess Readiness: Understand your current capabilities and set realistic expectations. Engage external partners

26. Pick SMART: For example, low hanging fruit of business problems that can be tackled via data or AI Data Availability and quality Business impact Feasibility Risk assessment Pick appropriate AI technique Begin with POC/pilot projects to validate concepts and demonstrate value. Pick first problem for POC Implement POC

27. Monitor and Iterate: Regularly review progress, adapt strategies, and scale successful initiatives.

28. Scale AI initiatives, upgrade infra

29. What to expect in future from AI? AI models improve quality, consistency, and reliability Focus on sovereign AI (using open sourced models) The focus on responsible AI increases Cost of AI models goes down over time ( commoditization ) More capable AI agents, integrations Replacing popular and expensive B2B SaaS Specialization of AI tools (branching)

30. Overrated or underrated? Short term: over rated Long term: underrated

31. ‹#› "Computers are useless. They can only give you answers." Pablo Picasso

32. What we talked about What AI? Some paths toward failure with AI A Framework for success, examples Future Recap

33. Questions? Stay in touch opcom.io Opcom GmbH Switzerland [email protected] msolujic