AI Case Study
Improved Safety and Compliance using AI-Powered Document Management for Oilfield Chemical Services
Smartbridge deployed a multi-agent AI assistant on Microsoft Copilot Studio, giving a Texas-based specialty oilfield chemical services company instant, citation-accurate access to Safety Data Sheets, operating procedures, HSE policies, lab protocols, and more on any mobile device, with no Microsoft Teams license required.
Our client is a privately held specialty oilfield chemical services company that supplies production chemicals, water treatment solutions, drilling and stimulation fluids, and field services. They are supported by a large, primarily field-based workforce of drivers, technicians, and service personnel.
In chemical production and supply, document access is a safety-critical operation. Field workers are legally required to produce Safety Data Sheets on demand at every point of chemical use. Operators must follow exact written procedures before performing hazardous tasks, and lab technicians must apply precise, documented test methods. A wrong answer, a vague summary, or a delayed response carries real consequences: regulatory exposure, a safety incident, or a quality failure.
For a company managing 500+ employees, thousands of products, and multiple document domains (all supporting a predominantly mobile, field-based workforce) the pressure to modernize document access was both operational and compliance-driven.
Key Operational Gaps:
Scale and Accuracy in Conflict
Impact Weight: 5 – Critical (Safety & Compliance)
Problem
The company maintained over 3,000 documents spanning Safety Data Sheets, HSE policies, DOT compliance, operating procedures, lab protocols, HR, and IT support. Consolidating all content into a single AI knowledge source produced slow, inaccurate responses. The search space was too large and too diverse. Accuracy degraded precisely where it mattered most.
Solution & Outcome
We partitioned knowledge by domain. Each specialist child agent indexes a single, scoped SharePoint folder and answers one category of question — making it a true expert in its domain rather than a generalist across thousands of unrelated documents.
A Mobile-First, License-Light Workforce
Impact Weight: 4 – High (Adoption & Accessibility)
Problem
The majority of the workforce such as field drivers, technicians, and service personnel operate without desktop computers or Microsoft Teams licenses. Any solution had to work on a personal or company mobile phone using only a standard Microsoft 365 account.
Solution & Outcome
The chatbot was deployed via the Microsoft 365 App channel, accessible on iOS and Android at no additional per-user license cost. Field workers can access the full chatbot experience, including Safety Data Sheet retrieval, from their phone, with no Teams license required.
Zero Tolerance for Document Approximation
Impact Weight: 5 – Critical (Regulatory & Safety)
Problem
Safety Data Sheets must be returned as source documents with full citations, not summarized or paraphrased. Summaries can omit or misrepresent hazard information. The same applies to operating procedures: the exact written procedure must be returned, not an interpretation.
Solution & Outcome
Parent agent instructions enforce strict pass-through rules: child agent responses are relayed to the user exactly as retrieved, with all source links and citations preserved. Summarization is explicitly prohibited in the agent’s configuration.
Operational Reality Before AI Modernization
Before the Copilot Studio solution deployment, field employees often relied on manual and inconsistent methods to retrieve operational documents. Safety Data Sheets, operating procedures, and compliance documents were stored across multiple SharePoint libraries and folders, requiring workers to search manually, contact office personnel, or wait for supervisors to provide the correct documentation.
For field technicians operating at customer sites, even small retrieval delays created operational friction. A worker needing an SDS for a chemical transfer, a DOT procedure for transport compliance, or an HSE protocol before performing work could spend valuable time navigating document repositories on a mobile device or calling dispatch and office staff for assistance.
The impact extended beyond productivity. Inconsistent access methods increased the risk of:
The company needed a system that could provide immediate, accurate, and mobile-first retrieval of regulated operational documents without requiring users to understand document locations, folder structures, or SharePoint navigation.
AI Governance and Safety Controls
Because the solution operates in safety-sensitive and compliance-driven workflows, Smartbridge designed the architecture with strict controls to minimize hallucination risk and preserve document fidelity.
Several safeguards were implemented to ensure reliable and auditable responses:
Source-Grounded Responses Only
The agents are configured to return information exclusively from approved SharePoint knowledge sources. Responses are grounded in existing enterprise documents rather than open-ended generative behavior.
Citation Preservation
All responses preserve original citations and source links, allowing employees to validate information directly against the underlying document.
No Summarization for Regulated Content
For Safety Data Sheets, operating procedures, and regulatory documentation, summarization behavior was explicitly disabled. Documents are returned verbatim to prevent omission or reinterpretation of critical safety information.
Domain Isolation
Each child agent operates within a narrowly scoped document domain, reducing retrieval ambiguity and improving answer accuracy.
Existing Security Model Retained
The solution continues to inherit existing Microsoft 365 and SharePoint security controls, ensuring users only access content permitted under enterprise identity and access policies.
Operational Monitoring
Azure Log Analytics provides centralized visibility into conversation activity, routing behavior, and system health, enabling ongoing monitoring and tuning of response quality over time.
Human Judgment Remains in the Loop
The chatbot supports operational decision-making but does not replace employee training, field verification, or required safety procedures.
Example Field Workflow
A field technician preparing for a chemical transfer at a customer site can now retrieve the exact Safety Data Sheet from a mobile phone in seconds by asking a natural-language question within the Microsoft 365 app.
Previously, the same process often required:
The new experience reduces retrieval friction while improving confidence that employees are using the correct, current, and compliance-approved documentation in field operations.
Measurable Impact Across Productivity, Compliance, and Field Access
The following impact estimates are derived from a productivity and risk model based on deployment data and oilfield services benchmarks. All figures are intentionally conservative.
Productivity Impact
| Agent Domain | Knowledge Scope |
|---|---|
| Safety Data Sheets | Seven scoped agents by product category – 100 to 500 documents each |
| HSE & Regulatory Compliance | Health, safety, & environmental policies; DOT compliance; operating procedures |
| Laboratory Protocols | Lab testing procedures and quality analysis methods |
| HR & IT Support | HR policies and employee guidelines; IT FAQs and application support. |
Compliance Risk & Access
Model assumptions: 500 employees, 60% field-based; 18 minutes saved per document query vs. manual SharePoint retrieval or office call; blended fully-loaded labor rate of $45/hr; 0.5% of annual revenue estimated at risk from incorrect document retrieval; chatbot reduces that exposure by 60%, with human judgment remaining in the loop.
Modernizing Knowledge Access for Safety-Critical Work
By combining Microsoft Copilot Studio with a domain-specialized multi-agent architecture, Smartbridge helped the organization transform document access from a manual administrative process into a scalable operational capability.
The solution improved mobile accessibility, strengthened document retrieval accuracy, reduced operational friction, and enabled field personnel to access critical safety and compliance information in seconds, all without increasing licensing complexity across the workforce.
For organizations operating in safety-critical industries, the project demonstrates how AI can deliver measurable productivity gains while preserving governance, compliance, and operational control.







