# Enterprise efficiencies with ## Nir Manor・Transformational Retail 15 April 2025 --- ## AI Evolution # AI waves **Agentic AI** Enables to manage and control reality. **Generative AI** Create and interact with reality. **Traditional AI** Allowed analysis and predictions. --- AI fundamentals ### Essential attributes of AI Agents - Continuously modifies plans in response to evolving circumstances to fulfill processes efficiently. #### Adaptable Planning #### Autonomy - Performs objective-oriented tasks with limited human supervision. #### Context Understanding #### Reasoning - Interprets and responds to natural language and other modalities. - Situation-aware decision processes, discretionary choices & compromise considerations. #### Action Enabled - Authorized to implement actions through integration with online platforms providing competencies. --- ## 10 core GenAI Use Cases for Retail/CPGs in 2025 - Redefining and accelerating product development - Optimize operations - Products descriptions and visuals - Democratize data access and analysis - Agentic AI for various enterprise tasks - Demand Forecast 3.0 - Retail planning optimization - Improve Customer Experience - Personalization at scale - New research trends - Strengthen working teams --- ### Examples #### Enterprise automation: AI-Agents examples - **Compliance & Risk Agent** - Monitors and ensures compliance with industry regulations and internal policies across all systems and workflows. - **Integration Coordinator Agent** - Manages and streamlines integrations across various enterprise systems like ERP, CRM, SaaS, cloud, on-premise, and legacy platforms. - **Sales Pipeline Agent** - Manages the sales pipeline by analyzing leads, predicting conversions, and automating follow-ups. - **HR & Workforce Agent** - Automates HR tasks like recruitment, onboarding, payroll processing, time-off management, and performance evaluations. - **Financial Controller Agent** - Automates financial tasks such as budgeting, forecasting, reporting, and revenue recognition by integrating data from multiple financial systems. --- ### Challenges AI Agents will be everywhere, but the challenges are: - The challenge is to seamlessly connect to the enterprise backend, which mostly has legacy and/or on-prem systems. - AI agents to act as per corporate policy related to the subject in case. - Complicated integrations to ERP, billing systems, and proprietary databases pose significant hurdles for AI agents. --- ### Innovation #### AI-Powered Financial Operations Unlocking new possibilities --- ### AI Agent orchestration – Use Cases - Automate data flows, validation, enrichment across multiple SaaS & internal databases. - **Data Enrichment & Automation** - Procure-to-Pay (P2P) - Order-to-Cash (O2C) - Automation of procurement workflows, connecting ERP (NetSuite, SAP, Priority) to procurement platforms (ZIP, Coupa). - Orchestrate end-to-end sales-to-finance process between CRM (Salesforce, HubSpot) and ERP systems. - **Customer Support** - Smart Data Mapping - Automation for Zendesk, ServiceNow, Freshdesk to ensure proactive CS. - Helping SaaS companies integrate with ERP/CRM systems of their clients. --- ### Challenges #### Procurement & cash management challenges - More purchasing autonomy by numerous departments - Countless vendors - Overspending (mainly before issuing a PO) - Cross-team approvals and risk reviews - Unclear ordering process and law orchestration --- ### AI Applications #### Procure-to-pay (P2P) – Use Case - Automation of Connecting ERP (NetSuite, SAP, Priority) to procurement platforms workflows (ZIP, Coupa). | Automation of procurement workflows | Connecting ERP (NetSuite, SAP, Priority) to procurement platforms(ZIP, Coupa). | | --- | --- | | Vendor performance analysis | AI Agents analyze historical vendor performance, pricing trends,and order urgency to recommend or automatically select the best supplier for each order. | | Invoice discrepancy resolution | Flags discrepancies when invoices don't match purchase orders or delivery receipts,classifies the issue,and resolves it or escalates to procurement. | | Inventory shortage prediction | Predicts shortages based on sales trends,supplier lead times,and seasonal fluctuations,triggering POs before shortages occur. | --- ### Case Study #### Walmart Case Study – P2P **AI deployment in** Walmart deployed an AI negotiation agent in its **procurement** Procure-to-Pay process, transforming how supplier contracts are managed. **Challenges in** About 20% of suppliers were on standard, non-negotiated **procurement** contracts, and most processes were manual, slow, and prone to mistakes. **Achievements and** The AI negotiator closed deals with 68% of suppliers, **savings** achieving 3% cost savings and extending payment terms by 35 days, leading to millions in savings. --- #### Order-to-Cash (O2C) - Use Case - **Contract Scanning for Compliance** - AI scans new customer contracts to identify missing details, inconsistencies, or compliance risks before they are approved. - **Automatic Payment Term Adjustments** - AI Agents analyze past payment behavior, market conditions, and risk scores to auto-adjust payment terms before a new contract is signed. - **Self-Learning Invoice Matching** - AI Agents match invoices to contracts and deliveries, using NLP-based document processing, auto-fixing common discrepancies. - **Predictive Payment Follow-Ups** - AI predicts late payments based on customer behavior and adjusts outreach strategies dynamically. --- ### Case Study #### Siemens Case Study – O2C #### AI in cash collection Pain points before AI Siemens uses AI for collection of cash from thousands of clients worldwide. - High write-offs rate, delayed payments causing cash flow problems, and heavy manual workload on financial teams. #### AI-Driven automation Results achieved - Dynamic risk assessment, reconciliation of documents, and personalized payment follow-ups have been automated. - 20% reduction in delayed payments, 10% reduction in write-offs, improved cash flow predictability, 25% reduction in manual work. --- ### Comparison #### AI vs. rule-based automation What differentiates them? - **Dynamic decision-making** - AI adjusts credit risk assessments in real-time. It responds to changing payment behaviors before problems occur. - Rule-based systems rely on static thresholds. They can't adapt to evolving situations. - **Natural Language Processing (NLP)** - AI extracts key points from unstructured text. It catches discrepancies in contracts that algorithms miss. - **Predictive analytics** - AI predicts payment behavior and prioritizes accordingly. - Personalizes collection strategies for each customer. --- ### AI in E-Commerce #### Product Description - Use Case - **Automated Content Generation** - Automatically generate, optimize, and personalize product descriptions, images, and category content across e-commerce platforms and digital catalogs. - **Natural Language Generation (NLG)** - Based on product specs, AI creates SEO-optimized and customer-tailored descriptions in multiple languages. - Continuously test variants of titles and descriptions to improve conversion rates. - Automatically generate lifestyle images or product mockups using product metadata. - **Visual AI Tools** - **A/B Testing** - **Omnichannel Adaptation** - Tailors messaging for specific channels (e.g., mobile, desktop, social media) using customer behavior insights. - Less manual workload, shorter T2M for new products, enhanced customer engagement, CVRs, ensured consistency and personalization across all digital touchpoints. --- ### Case Study #### The Home Depot Case Study – Product Descriptions **Challenge** Managing content for over 2 million SKUs. Ensuring consistency, SEO performance, and personalization at scale. **AI Use** Create and optimize product descriptions, images, titles, and long-form content. Tailored content to customer personas using behavioral insights. A/B tested variants to maximize conversion rates. **Results Achieved** Significant reduction in manual work. Increased conversion rates on product pages. Improved organic search visibility. Faster product onboarding with reduced manual workload. Consistent, personalized messaging across channels. --- ### Challenges #### Customer Service challenges 1. Manual and hard to scale 2. Long SLAs that reduce CSAT 3. Bad reviews that damage brand reputation 4. Hard to train with high employee churn --- ### AI for Customer Service NLP/LLM, integrations, personalized process Hey Jane, I'm sorry to hear the size of the boots from order #24345 did not work out for your husband. I've gone ahead and started the exchange process for new Legend | Black boots in a size 9. Please see the Happy Returns shipping label here. … and please send happy birthday to your husband! Getting info from multiple integrations Taking actions using APIs Exchange Legend Boots size 8.5 9 Issue a return label --- ### CSAT #### AI Agent’s CSAT vs. Human Agent’s --- **“** ## You can’t stop the waves, but you can learn to surf **Jon Kabat-Zinn** --- Scan & let’s connect on LI # Thank you! ## Nir Manor・Transformational Retail