Tornatech Accelerates Access to Business Data with Amazon Bedrock-Driven Agentic Assistant
Tornatech needed faster, easier access to ERP data such as inventory, work orders, and invoices to improve decision-making, speed up customer response, and reduce the burden on IT.
Challenge
When Routine Questions Took Hours to Answer
Tornatech, a worldwide leader in the design and manufacturing of fire pump controllers, set out to improve how employees and customer service agents access business-critical data across its enterprise resource planning (ERP) and document systems.
With standard operating procedures (SOPs) spread across several platforms and many teams relying on IT for ad-hoc Excel reports, finding the right information often took longer than needed. This sometimes slowed customer response times, increased reporting workload for IT, and made cross-department collaboration more difficult.
Employees often needed answers to routine questions such as:
- “What’s the ship date for this work order?”
- “Do we have enough of this component in stock?”
- “Was this invoice processed?”
- “What’s the current price of this product?”
- “What safety measures do I need to enter the R&D lab?”
Recognizing the need for faster, easier access to accurate data, Tornatech looked for a smarter and more streamlined approach.
Solution
An AI Chatbot Integrated with ERP
Partnering with Adastra, Tornatech implemented an AI-powered agentic chatbot built on Amazon Bedrock with features such as a text-to-SQL module that allows employees to interact with ERP and document systems in plain language.
Employees can ask questions such as “What is the ship date for WO-12783?” or “Show current stock for controller model X,” and the assistant’s structured query agent converts the request into a secure SQL query, parses Tornatech’s ERP and delivers an accurate answer within seconds.
The chatbot connects to Tornatech’s ERP through a secure, read-only layer that provides access to inventory levels, work orders, production tracking, sales figures, financial data, bills of materials, and pricing information. It also utilizes other agents to index purchase-order documents and SOP PDFs so employees can quickly retrieve relevant policies, safety instructions, or expense-report rules without searching across multiple systems.
Accessible to all staff directly within Microsoft Teams, the solution was powered with a single, low-cost shared license through MS Copilot Studio, with no additional costs. Adastra delivered the first working version within weeks and ensured a secure architecture by connecting only to read-only database instances.
Impact
From Hours to Seconds: Faster Insights for Everyone
The chatbot has improved the way Tornatech’s teams access and use business-critical information. It has streamlined daily operations, reduced the reporting burden on IT, and enabled faster, more reliable service for customers.
The chatbot changed how staff access and use information:
- Hours to seconds: Supervisors, after-sales representatives, and service agents now get instant answers during live calls or on the shop floor.
- Higher productivity: Eliminates repetitive IT-built Excel reports as staff now have self-serve insights on inventory, pricing, production, and customer orders.
- Trusted answers: Provides real-time, accurate data directly from ERP and SOP repositories.
- Improved customer experience: Enables faster responses on product availability, component reuse, work-order tracking, and invoicing.
- Executive support: The success inspired new initiatives such as applying a retrieval-augmented generation (RAG) system to more than 20 years of historical sales and service data for advanced after-sales support.
