At a Glance
- Industry: Manufacturing & Product Development
- Challenge: Manual risk assessment process requiring teams to start from scratch each time, limiting how quickly engineering capacity could scale across growing product portfolios
- Solution: AI-powered risk assessment agent that systematizes institutional knowledge for comprehensive failure analysis
- Key Result: 100% knowledge coverage in every assessment, reducing product risk and accelerating development cycles
The Challenge: Manual Risk Assessment at Scale
For product development teams in regulated industries, risk assessment isn't optional. Every component must be evaluated for potential failure. Every failure mode needs documentation. Every risk requires a priority score before a product ships.
At DEMA Engineering Company, a leading supplier of chemical dispensing systems and industrial valves, the operations team knew this reality well. Their Design Failure Mode and Effects Analysis (DFMEA) process was thorough, rigorous, and time-consuming.
Each new product triggered the same cycle: 6-8 meetings with 8-10 engineers gathered around a conference table, manually brainstorming how every component could fail. Behind the scenes, relying on manual recall to compile this analysis into living spreadsheets left the process vulnerable to missed edge cases and delayed product pipelines.
"We were essentially starting from scratch every time," the operations lead explained during discovery. "We had decades of institutional knowledge, but it lived in people's heads and in past DFMEA documents scattered across SharePoint."
The emotional cost was real. Teams were manually reconstructing analyses that already existed in historical documentation. And the organization's most valuable asset, accumulated engineering expertise, wasn't organized into a system that could accelerate future product development. DEMA needed to find a solution that expanded engineering capacity for product portfolio growth.
The Solution: Designing an AI Risk Assessment System
The SoftSnow team started by understanding the process. This involved conducting multiple discovery sessions, mapping the existing workflow, and analyzing the data structure. DEMA’s operations team shared historical DFMEA documents, and we began organizing them into product-specific databases stored in their SharePoint.
SoftSnow selected Cassidy AI, a third-party work automation platform, as the orchestration engine for this solution. It connects to existing systems like SharePoint, reads company data in context, and executes multi-step workflows that produce structured outputs.
For DEMA, this meant the agent could pull from their historical knowledge base, process complex Excel files, apply their specific risk assessment logic, and deliver completed documents without manual intervention. The databases sync automatically from SharePoint to Cassidy, ensuring the system always references the most current institutional knowledge.
The DFMEA Risk Assessment Agent orchestrates multiple steps to turn institutional knowledge into a working system. Here's the flow:
- The team provides three inputs: Product name, product type, and an Excel file containing stakeholder, feature element, and component element columns.
- When the workflow runs, Cassidy reads the input file row by row. For each component element, the agent searches the relevant product database for similar entries, looking for patterns across feature descriptions, component types, and historical failure modes to pull the most relevant examples.
- Once the agent identifies relevant historical data, it populates the new spreadsheet with potential failure modes, failure effects, causes, and estimated RPN values. The output matches DEMA's existing DFMEA template format.
- The workflow delivers the completed Excel file with the results via email.
Engineers now start with a comprehensive failure analysis drawn from the organization's complete expertise. What used to be a manual bottleneck now runs automatically, freeing the team to make faster, better-informed safety decisions and focus on edge case evaluation.
"It gets us 80% of the way there," the operations lead said during the demo. "We still review everything, adjust language, validate the risk scores. But instead of staring at a blank spreadsheet, we're refining analysis that's already grounded in what we've learned."
The Results: Scalable Engineering Intelligence
As the Risk Assessment Agent was implemented, the nature of the work changed. Engineers now arrive at meetings with pre-populated analysis instead of blank templates. Conversations shift from "What could go wrong?" to "Is this assessment accurate?" The team catches edge cases faster because they're reviewing comprehensive lists.
The agent also creates a feedback loop. As new products get assessed and new failure modes emerge, the team adds them to the SharePoint databases. The agent automatically incorporates that knowledge into future assessments within 24 hours.
"We're not just saving time," the operations lead reflected during training. "We're designing a system that gets smarter with every product we develop."
The business impact:
- 100% institutional knowledge coverage in every assessment
- Accelerated product cycles by eliminating safety review bottlenecks
- A continuous learning system that improves with every product
- More consistent risk identification across all products
Why This Matters for Your Product Development Process
Engineering teams accumulate tremendous expertise through years of product development. That knowledge represents a competitive advantage if you can systematize it.
Risk assessment processes exist across industries. Compliance documentation. Quality assurance workflows. Testing protocols. Any process that depends on institutional knowledge and repetitive analysis faces the same challenge that DEMA solved.
Want to explore how AI agents could transform your product development workflows? SoftSnow helps teams turn repetitive processes into scalable systems. We start by understanding your engineering expertise, organizing your historical data, and designing agents that amplify what your team already does well.
Let's talk about your specific challenges and how systematic intelligence could expand your team’s capacity.
