Peak at AWS re:Invent 2019
- 2. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved. Take AI/ML from theory to practice with Intel technologies on AWS A I M 2 0 3 - S Binay Ackalloor Director Intel AI Business Development for Cloud Service Providers
- 3. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved. Spotlight on: Delivering connected retail with AI A I M 2 0 3 - S Richard Potter CEO Peak Tom Summerfield Head of Commerce Footasylum
- 4. Agenda 1. Retail landscape 2. Connected retail 3. The AI tech stack for retail 4. AI Infrastructure 5. AI Studio 6. Peak solutions 7. Q&A
- 6. Its not all doom and gloom +4% YoY growth +3% YoY growth US retail sales UK retail sales *Source: Securities and Exchange Commission (SEC)*Source: ONS
- 8. We are Footasylum Creativity connects our bricks, clicks and kicks. • Through our 60+ stores, website and own brands we’re story tellers, intent on pushing boundaries. • We became a part of our consumers’ community: urban style leaders who drive trends and do things our own way. • Endorsed by this community of tribes we’re young and agile – never straight laced.
- 9. Retailers need to be better connected Context ControlConvenience With customers
- 10. Retailers need to be better connected Context ControlConvenience …and internally too. Marketing Supply chainMerchandising
- 11. Connected retail isn’t easy Understand your consumer Ensure efficient delivery Understand product interactions 1m+ consumers 60+ stores 25k products 3x commerce channels Omnichannel Fulfilment
- 12. AI supports connected retail Makes sense of data Supports the three Cs and puts consumer first Connects data, workflows and functions. Optimizes supply and demand
- 13. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
- 14. Built to deliver outcomes
- 15. AI: From hype to reality Computational power Scalable cloud infrastructure AI System and solutions The data to operationalize AI Connected retail experience
- 16. Peak’s AI System gives retailers everything they need to succeed with AI, accessed via customizable web-based apps and system integrations
- 17. The Peak AI System The full-stack AI System, built to deliver AI-driven outcomes across an enterprise. An intelligence layer across the enterprise
- 19. Store and unify data in our secure and scalable infrastructure
- 20. Configure AI solutions to your business’ needs, or build custom ones
- 21. Activate Peak solutions, either standalone or interconnected, and send outputs directly into business systems using APIs
- 22. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
- 23. Peak’s AI Infrastructure allows the AI System to seamlessly integrate with a wide-range of connectors. Seamless integration
- 24. Data ingestion and processing at scale Real time streaming of enterprise datasets Apache Spark on EMR for processing Stored on S3 and Redshift Data availability on time for multiple workflows
- 25. MLOps in a box Cloud-native, built for AI and ML Support for open-source tools and frameworks Several instance types to fit various ML use cases
- 26. Enterprise-grade security and control Enterprise-grade security and compliance Built-in controls for identity, data and network access for assets Governance with audit trail and resource management
- 27. AI Lifecycle Discovery Scope & proposal Build & Configure Deploy & Run Optimization & Management • Exploration • Prioritization • Feasibility • Business KPIs • Data exploration • Data preparation • Build & train • Evaluate • Deploy • Integrate • Monitor • Evaluate • Validate KPIs • Support • Re-training
- 28. Significant resource needed Discovery Scope & proposal Build & Configure Deploy & Run Optimization & Management • Exploration • Prioritization • Feasibility • Business KPIs • Data exploration • Data preparation • Build & train • Evaluate • Deploy • Integrate • Monitor & Evaluate • Validate KPIs • Support • Re-training Resource requirements: Data Sc. Data Eng. SW Eng. AI Ops. Biz Ops. Data Sc. Data Eng. SW Eng. AI Ops. Biz Ops. AI Ops. Data Sc. Biz Ops.
- 29. Significant technology needed Discovery Scope & proposal Build & Configure Deploy & Run Optimization & Management • Exploration • Prioritization • Feasibility • Business KPIs • Data exploration • Data preparation • Build & train • Evaluate • Deploy • Integrate • Monitor & Evaluate • Validate KPIs • Support • Re-training Platforms: • ETL/ELT • Data Streaming tool • Data Warehouse • In-memory database • Data Wrangler • ML Workbench • Continuous Integration • ML pipeline • In-memory database • Container-orchestration • ML Workbench • API Monitoring tool • In-memory database • Container-orchestration system
- 30. …not with Peak! Discovery Scope & proposal Build & Configure Deploy & Run Optimization & Management • Exploration • Prioritization • Feasibility • Business KPIs • Data exploration • Data preparation • Build & train • Evaluate • Deploy • Integrate • Monitor & Evaluate • Validate KPIs • Support • Re-training Full stack AI system:
- 31. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
- 33. Customizable UI and integration
- 34. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
- 35. Linking ML tech with retail needs Context Convenience Hyper-personalized marketing comms Recommendations and targeted promos AI powers buying decisions Warehouse & labor optimization Customer expectations Control
- 36. Customer AI Get more from your marketing CONTEXT Message personalization Recommendation engine AI driven Social media targeting HYPER-PERSONALIZATION = SALES Higher return on ad spend Increased conversion and higher AOV Reduced churn, increased campaign revenues PREDICTION Customer data Predictive Customer View +
- 37. CONTEXT 28%Uplift in email revenue 10xReturn on Ad spend (ROAS) Rapid time to value
- 38. Demand AI CONVENIENCE AI-driven demand planning and optimization IMPROVE PROFITABILITY Increased return on capital Improved efficiency and CX Higher margins, faster sell through PREDICTION Customer data Predictive Demand View + = Markdowns and promotions Inventory allocation Buying optimization OPTIMIZATION Assortment
- 39. Demand AI in practice CONVENIENCE Demand modelling suggests optimum prices for products This maximizes margin, balancing rate of sale with markdown size Most pricing decisions made automatically, increasing time for creativity in merchandising
- 40. Supply AI Fulfill demand at reduced costs CONTROL Logistics optimization Workforce planning Warehouse optimization OPTIMIZATION = IMPROVE PROFITABILITY Reduced stock movement costs Minimized labor costs Increased fulfilment efficiency PREDICTION Customer data Predictive Supply View +
- 41. Supply AI in practice CONTROL AI System determines optimum stock levels Supply AI connects to Demand and Customer AI Stock allocated intelligently across pick- face and bulk storage
- 42. Delivering connected retail Amazing customer experience Optimizing for profit
- 43. Delivering connected retail Amazing customer experience Optimizing for profit The future of retail. Driven by AI
- 45. THE only AI stack Scalable + Secure + Outcome-focused Focused on what’s important for future-facing retailers. Peak.ai/reinvent Visit us on the Intel stand and find out what your AI opportunity could be
- 46. © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
- 47. Thank you! © 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved. Richard Potter Richard.Potter@peak.ai Tom Summerfield Tom.Summerfield@footasylum.com