Get AI Ready - Laying the Foundations for Success

Slide deck from the fourth in BrightGen's Get AI Ready! webinar series. Presented by Keir Bowden, CTO BrightGen, with special guest Marius Rubin, CTO Credera. In this episode we take a step back from the laser focus on Salesforce to consider how to set yourself up for success across the wider enterprise.

Transcript

1. GET AI READY! www.brightgen.com www.credera.co.uk Part of Credera and the Omnicom Group of companies LAYING THE FOUNDATIONS FOR SUCCESS 29th November 2023
2. Agenda • Introductions • Enablers for success • Cloud • Data • Engineering • Risk • Q&A
3. Meet your speakers Keir Bowden Chief Technical Officer • 15 years at BrightGen • 36 years in the industry • Salesforce Certified Technical Architect • Salesforce MVP Hall of Fame
4. AWARD-WINNING SALESFORCE IMPLEMENTATION PARTNER 2023 Outstanding implementation partner We work across all sectors, business functions and company size Our service is flexible to the needs of our customers Capability to support customers from strategy to execution at a global scale We leverage the breadth of the Salesforce platform, have multi-cloud experience and are one of the most certified Service Cloud partners Strategy & advisory 01 Technology consulting & implementation 02 Enhancements & service management 03 500+ Salesforce Certifications
5. Meet your speakers Marius Rubin Chief Technology Officer • 9 years at Credera • Hands-on work, up to CIO advisory across public and private sector… • And covering AI, cloud, big data, software
6. Credera is a global boutique consulting firm that helps leading brands unlock extraordinary results REPRESENTATIVE RECOGNITION IN NUMBERS Employees Globally 4000+ Established in the UK 1989 ATLANTA CHICAGO COSTA RICA DALLAS (HQ) DENVER BERLIN LONDON MANCHESTER NEWCASTLE LEEDS AUCKLAND HYDERABAD SINGAPORE SYDNEY HOUSTON LOS ANGELES NEW YORK SAN FRANCISCO SEATTLE AMERICAS APAC EUROPE CLOUD Harness benefits of the cloud DATA & AI Exploit the power of your data ENGINEERING Production-ready without compromise TRANSFORMATION Ignite business outcomes STRATEGY Create competitive advantage MARTECH Cutting edge tech LOCATIONS Client Satisfaction 8.6 6
7. 7 OUR CREDENTIALS AI already represents 20% of Credera’s business and is our fastest growing service area O T H E R C L I E N T S W I T H W H O M W E ’ V E D O N E A I W O R K HEALTHCARE PUBLIC SERVICES FINANCIAL SERVICES
8. Strategy & Roadmap Data Technology Adoption & Scaling Delivery Model Talent Governance Compliance Privacy Transparency First Principles Risk BUSINESS VALUE AI: Enablers for success • AI enablement could be quite daunting and… • The implications across an organisation are huge but… • We’ve been here before. • We’re going to spend time talking about what technologists should be working on to make AI a success.
9. Cloud Salesforce Data Cloud Externally Hosted Services Model Studio External Model Feedback Your Custom Predictive and Generative Services Invoke Predictive or Generative AI AI to Actions • Flows • Recipes • Segmentation • Activation • Analytics
10. Cloud Scale & Scalability AI technologies are at their best when paired with your data at scale. They can also be very compute-hungry. Modularity Imagine you’re deploying multiple different models for different purposes or doing A/B testing. Proximity For user-facing systems you’ll want your data, your AI service and your web apps to be in one place. Rich capability For both hosted models & all the other stuff, it’s hard to match this capability anywhere else.
11. SYSTEMS OF ENGAGEMENT ARTIFICIAL INTELLIGENCE / MACHINE LEARNING CORE SYSTEMS OF RECORD EXISTING NEW CORE SYSTEMS / SYSTEMS OF RECORD HR SAP SUPPLY CHAIN ERP … APPLICATIONS INTERNAL APPLICATIONS EXTERNAL APPLICATIONS THIRD-PARTY APPLICATIONS Tools that improve productivity Tools that deliver differentiation Tools that deliver differentiation API MANAGEMENT PLATFORM ENGAGEMENT CHANNELS MOBILE TEAMS WEB OTHER ARTIFICIAL INTELLIGENCE / MACHINE LEARNING CORE ML CAPABILITIES DATA PRODUCT DATA STORAGE MODEL 2 GEN-AI CAPABILITIES DATA PRODUCT DATA STORAGE MODEL 1 TRANSPAENCY SECURITY & PRIVACY FAIRNESS RELIABILITY & SAFETY ACCOUNTABILITY INCLUSIVENESS MLOPS CAPABILITIES DATA PRODUCT DATA STORAGE MODEL 3 … CAPABILITIES DATA PRODUCT DATA STORAGE MODEL …. API MANAGEMENT
12. Data You’re not that likely to do much of this stuff You’ll probably find yourself doing this stuff 12 Using lots of compute training & clever stuff… A model (e.g. ChatGPT) Very smart people Lots of data… (think trillions of data points) Costing lots of $$$ User input Prompt Response Add rules/constraints (e.g. be concise) Return content AI-powered ‘product’ with access to business/user data etc. Apply access/data security controls 🚨 Beware! 🚨 This is a serious simplification. Even getting stuff from the data box to the product to the model requires bespoke engineering. Ethical and Copyright Considerations
13. Data Priority & Prep You’re going to have to make your data itself AI- ready. Quality & Governance A LLM won’t fix issues in your data - it will just make it more convincing. Differentiation The data you own is what differentiates you. A LLM will maximise its value. Platform In a heterogeneous estate, it’s critical – or you might have it all in one product already. We’ll get to Ops in a moment.
14. Engineering Automate all the things Remember LLMs (incl. 3rd party) are evolving rapidly, so you need systems that can change fast. Embrace ML/Data/LLMOps This means a robust environment strategy for your data users. Testing Your LLM solutions are more than ever ‘products’ and deeply integrated into production functionality Monitoring and human feedback These products will not behave consistently in production – certainly not over time. Further reading: § The Big Book of MLOps © databricks § The Path Towards ML-ightment © Credera
15. Risks
16. Risks Security Novel architectures present novel issues. And all the historic user data access challenges still apply. Legal/Technical AI services usually come with non-standard T&Cs. Even logging can present an issue! Cultural Transparency, fairness, inclusivity – and whether the AI capability reflects your organization.
17. Q&A