How MDS converted years of spreadsheet-locked maintenance data into natural language decision intelligence for a multinational specialty chemicals manufacturer — in 40 hours, on a fully Microsoft-funded roadmap.
Customer: A multinational specialty chemicals manufacturer Industry: Specialty Chemicals Manufacturing Employees: 350 Operations: Chile, Brazil, Germany Solution: A maintenance intelligence agent built on Microsoft Copilot Studio
The Customer
This organization is a 350-person specialty chemicals manufacturer producing construction chemical solutions, with operations spanning Chile, Brazil, and Germany. Technology investment here is driven by business necessity, not trend — leadership evaluates every investment against concrete operational value, with no budget for experiments that don’t produce measurable results. That pragmatic, business-first culture meant any AI initiative had to earn organizational trust before earning organizational investment.
How It Started: Agent in a Day
This is not a story of an inbound request scoped into a chatbot. MDS identified the organization as one with strong Microsoft 365 infrastructure and unleveraged potential for agentic AI, and brought its leadership into a Microsoft Agent in a Day workshop — a structured, hands-on discovery event, not a demo — before any commercial conversation began.
Through operational discussion, the customer’s own leaders surfaced maintenance operations as high-value and high-priority: years of machine histories, failure logs, and service data locked in one spreadsheet per machine, per month, per facility. Experienced technicians carried institutional knowledge that was inaccessible to newer team members and unavailable outside business hours. No pattern-based intelligence existed to prioritize servicing or identify recurring failures — the data was there, but the intelligence was not.
The workshop created the relationship and the mandate: leadership left aligned on a business-grounded use case before a single proposal was written.
A formal Microsoft Copilot Readiness Assessment across 28 capability dimensions followed, returning a 2-Star AI maturity score — with critical gaps identified in agentic automation (Very Low), conversational AI design (Very Low), cloud integration (Very Low), and data security (Low). No autonomous workflow capability existed. No AI governance structure was in place. No adoption framework had been defined. That assessment became the blueprint for every Microsoft-funded engagement that followed.
The Challenge
Maintenance operations across the organization’s facilities ran entirely on spreadsheets — one sheet per month, per machine, per facility. Years of machine histories, failure logs, technician notes, and service records were operationally invisible, inaccessible as structured intelligence. Maintenance personnel spent significant time manually reviewing historical files before making any servicing decision. Institutional knowledge was fragmented across individuals and files. No pattern-based system existed to prioritize equipment servicing or distinguish preventive from corrective maintenance needs.
The business goal was clear: convert accumulated maintenance data into AI-powered operational decision support — eliminating manual review time, improving prioritization accuracy, and demonstrating that Microsoft Copilot technologies could deliver measurable value inside a core manufacturing process.
The AI Advisory and Transformation Plan
MDS executed a sequenced, fully Microsoft-funded AI transformation program:
- Copilot Readiness Assessment — established an objective baseline, identified the three highest-priority capability gaps, and created the evidence foundation required to unlock Microsoft investment for subsequent engagements.
- Microsoft Agent in a Day Workshop — engaged technical and business stakeholders across Chile and Brazil through structured operational discovery that led the organization’s own teams to identify maintenance intelligence as the highest-value AI use case.
- Microsoft-funded Copilot and Power Envisioning and Proof of Concept — translated that use case into a production maintenance intelligence agent designed to address the specific agentic and data accessibility gaps the assessment had exposed.
- Microsoft-funded Data Security engagement (now active) — addressing the data security and cloud integration gaps diagnosed at the outset, preparing the organization’s data estate for responsible enterprise-wide AI expansion.
The Solution: MaintainIQ
MDS designed and delivered a production maintenance intelligence agent — internally named MaintainIQ — built on Microsoft Copilot Studio, Microsoft Dataverse, Power Apps, and Power BI. It was delivered in 40 hours across assessment, design, a sprint development session, and demo delivery — a pace possible only because readiness and envisioning had already resolved every ambiguity.
The technical build:
- ETL and data consolidation. The most technically complex element was the ETL process: historical maintenance data existed in individual monthly spreadsheets requiring consolidation, structural realignment, and transformation into a single queryable format readable by the RAG system. Multiple alignment sessions defined the correct data structure and tooling before development began.
- Prompt engineering. The agent’s natural language capability required over 7,000 characters of prompt engineering — structured definitions of operational concepts, column logic, and response behavior — enabling maintenance teams to ask unrestricted questions against the full maintenance history with no predefined query limitations.
- Dataverse migration. Historical records were migrated from unstructured monthly spreadsheets into Microsoft Dataverse, creating the organization’s first governed, queryable operational data layer. Preventive and corrective actions were distinguished through structured data entry, with AI further analyzing free-text comments to identify inconsistencies. This migration was specifically architected as the production path to eliminate performance bottlenecks at scale.
- Power Apps and Power BI. Power Apps provided the maintenance team interface and workflow integration. Power BI surfaced trend dashboards and equipment-level intelligence for operational decision-makers.
Trust: Security, Governance, and Responsible AI
MDS embedded responsible AI principles into the engagement from day one. During the proof of concept, maintenance files were stored in SharePoint with access restricted to authorized personnel, providing immediate access governance. In production, Microsoft Dataverse’s native permission framework enforces record-level data access controls.
Rather than proceeding to enterprise expansion without addressing the assessed security gaps, MDS sequenced a dedicated Microsoft-funded data security engagement — currently active — implementing Microsoft Purview for data classification and Microsoft Entra for access governance across the organization’s multinational operations.
Human oversight was preserved throughout: maintenance personnel receive AI-assisted recommendations grounded in governed Dataverse records, not autonomous commands. No unvetted external data sources are accessed.
Workflow Transformation
Before MDS: A maintenance technician manually reviewed multiple spreadsheet files, searched failure history by memory or folder navigation, and made prioritization decisions from incomplete, unsystematic data.
After MDS: The same technician queries the Copilot Studio maintenance agent in natural language — “Which machines are highest priority for servicing this week?” — and receives AI-generated recommendations grounded in the full historical maintenance record, with failure pattern analysis, maintenance frequency data, and action type classification surfaced automatically. Institutional maintenance knowledge that previously existed only in experienced technicians’ minds is now accessible to every authorized team member, consistently, at every shift, across every facility.
Measured Outcomes
| Metric | Result |
| AI Maturity to Production | 2-Star baseline → live production agent |
| Time to Production | 40 hours across assessment, design, sprint development, and demo |
| Prompt Engineering | 7,000+ characters enabling unrestricted natural language querying |
| Countries in Scope | 3 (Chile, Brazil, Germany) |
| Microsoft Platforms Activated | 4 (Copilot Studio, Dataverse, Power Apps, Power BI) |
| Manual Maintenance Analysis | Eliminated — AI-assisted recommendations replace spreadsheet review for all prioritization decisions |
| Executive Sponsorship | Secured across two countries within a single funded engagement cycle |
| Data Security Engagement | Active — deploying Microsoft Purview and Entra to govern the AI data foundation |
| Expansion Roadmap | Formal roadmap established across additional departments and geographies |
Why This Matters: A Repeatable Model
This is one of MDS’s most complete AI transformation engagements to date: a sequenced, fully Microsoft-funded program spanning a formal readiness assessment, a structured envisioning process, a production agentic AI deployment, and an active data security engagement preparing the organization for enterprise-wide AI expansion across three countries.
This is not a pilot or a demo. It is Phase 1 of a funded, multi-year transformation — and it started with a workshop.
The MDS framework applied here is fully repeatable across MDS’s Latin American manufacturing customer base: Readiness Assessment → Agent in a Day → Envisioning and PoC → Data Security Foundation → Enterprise Expansion. Each stage is anchored to a diagnosed gap, Microsoft-funded, and structured to expand platform consumption. A production-ready version of this maintenance intelligence agent — including the recommended architectural changes for Dataverse at scale — can be deployed for a new manufacturing customer in under 30 days. The governed Dataverse foundation, natural language interface, and pattern-based prioritization logic are directly transferable across facilities, countries, and manufacturing verticals.
Maureen Data Systems (MDS) is a New York-based Microsoft Solutions Partner delivering AI, Data, and Security transformation across manufacturing, financial services, government, and professional services. Learn more at www.mdsny.com.