video1.0<iframe src="https://www.loom.com/embed/52229ee9f2d74458980d392c6bf7b3d6" frameborder="0" width="2048" height="1536" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>15362048Loomhttps://www.loom.com15362048https://cdn.loom.com/sessions/thumbnails/52229ee9f2d74458980d392c6bf7b3d6-1720044083848.gif3797.12Leveraging AI for Responsible Sourcing: A Breakdown and Webinar FAQ 1. Introduction Webinar: Leveraging AI for responsible sourcing by Mosaic RSR. Hosts: Ernest and Greta Matos from Mosaic RSR. Focus: Responsible sourcing, social compliance, and AI technology. 2. Background of the Speakers Greta Matos: Advisor at Mosaic RSR with two decades of experience in responsible sourcing. Aparajita Thakker: Co-founder at Esger, specializing in AI technology for social impact. 3. Overview of Responsible Sourcing Process Social Compliance Programs: Initiated due to exposés on poor working conditions. Proliferation: Voluntary social compliance programs by brands and certification teams. Increase: Social compliance auditing to mitigate risks in supply chains. Challenges: Standardized audits, redundancy, and audit quality. 4. Opportunities for Improvement Shift Focus: From assessment to supplier development for localized improvement. Challenges: Lack of time, energy, and resources for supplier development. Regulatory Requirements: Introduction of mandatory human rights due diligence laws. 5. Role of AI in Responsible Sourcing Automation: Processes like risk assessments, audit reviews, and corrective action plans. Capabilities: Recognizing speech, decision-making, and identifying patterns. Benefits: Saving time, improving efficiency, and providing actionable insights for supplier improvement. 6. Use Cases of AI in Responsible Sourcing Assessing Impact: Audit reviewing, tracking supplier performance, and CAP management. Automation: Risk scoring, data analysis, equivalency checks, and CAP generation. Ensuring Compliance: Accuracy, confidentiality, and adherence to local and international laws. 7. Mosaic and Esger Collaboration Development: AI-powered tools like AuditSense for flagging non-conformances. AI Capabilities: Assigning risk scores, generating corrective action plans, and providing expert insights. Data Analysis: Analyzing data from various sources, providing supplier risk profiles, and actionable insights. 8. Closing Remarks and Q&A Demo: AuditSense AI platform showcasing risk scoring, CAP generation, and insights dashboard. Discussion: Opportunity for questions, further discussions, and potential pilot programs. Emphasis: Potential of AI to revolutionize responsible sourcing and social compliance. Webinar Question & Answer: Q: Is the premise here that customers will develop CAP for their suppliers? Or will suppliers develop their own CAP? A: The premise discussed during the webinar is three-fold: The AI performs an equivalency review of different audit reports and generates a client-specific CAP for the suppliers. The AI generates CAP responses based on non-conformances identified, and these can be shared directly with suppliers or brands wishing to share CAPs with their suppliers. Mosaic and Esger are exploring how AI can play a more resource-oriented role in the CAP review process, streamlining the actual review while providing suppliers with supportive insights about improvement resources. This involves teaching the AI to provide feedback on the completeness of the CAP and quality of the supplier response. The supplier develops their own CAP, submits it for review, and the AI reviews the inputs and provides feedback on the completeness and quality of the supplier inputs, with a compliance expert maintaining oversight of this process. Q: AI models tend to need a lot of data to be trained. Does Esger’s AI use its own data or can our organization’s data be used to train it? If there’s not enough data, what can be done? A: Esger’s AI only uses individual organizations’ data for analysis (which can include audit reports, worker voice data, third-party data sources, or questionnaire data). The AI relies on available private or public data, and it cannot solve the problem of lacking data. Q: How does Esger ensure the CAPs created meet local and international law? A: Esger uses "Explainable AI," where each CAP is attributed to local laws or principles (such as the ETI principles). This attribution ensures compliance and that CAP recommendations are grounded in regulation. Q: How is the quality of content and validity of information evaluated with reduced time? A: The AI is trained on a baseline of ESG information and fine-tuned on specific company criteria and requirements, enabling it to flag risks and generate custom CAPs. The AI has often caught errors missed by manual review processes. Q: Can you talk about cost savings and the impact on Mosaic’s platform cost? A: Access to Tess within the Mosaic RSR platform is included in the standard subscription, so there is no direct impact on current pricing. The price-per-question model applies only to questions sent directly to Mosaic in-country experts. Q: What is the cost associated with implementing AI for different tasks? A: The cost depends on multiple factors, including data involved, regions, security requirements, and access control. Previous implementations have shown AI can reduce costs by up to 75% compared to manual methods for tasks like audit equivalency, risk scoring, CAP generation, and CAP review. Q: How does Esger ensure proprietary rights and data privacy in audit protocol comparisons? A: All data, including audit protocols, are encrypted over the internet, ensuring no unauthorized access. Esger does not save any data, preventing data breaches. Q: Does your AI fetch relevant local laws, and how is the information kept updated? A: Mosaic supports the process by training the AI on verified, relevant local law sources, international standards, and industry-specific best practices. Q: How can AI aid in remote auditing? A: AI can support the document review process, allowing auditors more time for site walkthroughs, worker interviews, and management interviews. This could fundamentally change audit approaches by enabling a better understanding of system gaps before the audit. Q: Can you edit CAP suggestions? A: Yes, once the CAP is generated, it can be modified before finalization. Q: How do you ensure the credibility of reports and filter out "bad" reports? A: The quality of data input correlates with the data output. AI can catch inconsistencies and flag quality concerns if the data input is inconsistent with program or certification scheme SOPs. AI can learn protocols to differentiate "good" reports from "bad" ones and apply this at scale.