BASF Coatings Deploys Marketmind Multi-Agent Supervisor System with Databricks for Sales & Marketing

Title

BASF Coatings Deploys Marketmind Multi-Agent Supervisor System with Databricks for Sales & Marketing

Content

BASF Coatings, the coatings division of German multinational BASF specializing in automotive and industrial surface technologies, partnered with Databricks and Accenture to deploy a production-ready multi-agent supervisor solution aimed at automating coordination and analysis across cross-departmental digital workflows. The organization, with over 11,000 employees across more than 70 sites worldwide, adopted the supervisor agent architecture to handle the complexity of its data landscape — structured Delta tables managed by BASF's central Enterprise Data Lake organization alongside unstructured Salesforce visit reports and market news PDFs arriving at different speeds. The first production project, named Marketmind, focuses on the Sales & Marketing division and consolidates internal Salesforce customer visit reports and market consumption insights with external market trends including S&P 500 news. The Marketmind architecture is built on the supervisor pattern from LangGraph, coordinating specialist agents — specifically Databricks AI/BI Genie agents for structured data and function-calling agents for unstructured retrieval — which interact with the Databricks Vector Store Retrieval tool. AI/BI Genie uses metadata from Unity Catalog (table descriptions, PK/FK relationships, column descriptions) to parse user questions, construct SQL, and deliver contextually relevant answers, mitigating errors and hallucinations. Genie authors enhanced the space with curated tables, join relationships, value sampling, SQL examples, and BASF-specific instructions, while regular Benchmark tests evaluated performance as new data arrived. For unstructured data such as Salesforce visit reports and market news, the team built vector search indices for each source using embeddings for context-aware similarity search, then wrapped Databricks AI Search queries in Unity Catalog functions to ensure enterprise-grade governance, discoverability, and automatic MLflow tracing. The deployed supervisor endpoint is integrated with Microsoft Teams for real-time agent execution, embedding conversational deployment endpoints directly within the Teams interface so that business stakeholders can interact with agents using natural language. The project began its scoping phase in April (2026), followed by a 5–6 week proof of concept, then full implementation accompanied by technical upskilling workshops, architecture reviews, and product/feature discussions with the Databricks product team. A one-month pilot with 25 key users preceded final refinement, with go-live and North American rollout planned by end of October — at which point more than 1,000 sales representatives worldwide will be using Marketmind. In production, Marketmind delivers personalized notifications alongside suggested actions and strategies based on current market events; sales reps can dig deeper into underlying data via an easy-to-use chat interface, with proactive adaptive cards surfacing the latest market trends weekly and a voting mechanism (thumbs up/down plus written feedback) feeding into the model inference table to improve agent quality. 'Marketmind turns our field interactions into timely, AI-driven actions — nudging smart follow-ups, surfacing relevant opportunities, and connecting peers facing similar challenges. The result: faster prep, sharper customer conversations, and more time selling where it counts,' said Adrian Fierro, Head of Global Market Intelligence at BASF Coatings. The deployment is built on Databricks' enterprise-grade governance capabilities including Unity Catalog for fine-grained access control, lineage tracking, and auditability.