Discover
Automating Quality
Automating Quality
Author: SOLABS, Mandy Gervasio, Philippe Gaudreau, and Guests
Subscribed: 11Played: 144Subscribe
Share
© 2018+ SOLABS ALL RIGHTS RESERVED
Description
Welcome to the Automating Quality show sponsored by SOLABS, with your host Mandy Gervasio, Technology and Life Sciences industry veteran. The Automating Quality podcast is designed to provide professionals in the regulated Life Sciences industry with best practice perspectives as well as employable strategies and tools relevant to current industry trends. Listeners will come to understand pressing issues in the space and hear best in class thought leadership on various topics such as Quality, Training and Regulatory Compliance driven from an automation lens.
69 Episodes
Reverse
Welcome to Automating Quality, the life sciences–focused show exploring how quality, risk, and technology intersect to modernize regulated environments. In Part 2 of this two-part series, Philippe Gaudreau continues the conversation with Dr. John Salvato, founder of Design Quality Services and an ISO 13485 lead auditor with more than 30 years of experience in medical device and automotive engineering. Having covered the foundations of design controls and FMEA in Part 1, the conversation now turns to where artificial intelligence fits into risk management and design documentation, and where it does not. Salvato distinguishes what AI can responsibly accelerate in FMEA from what it cannot. Once a methodology is structured around clear rubrics, he explains, AI can apply it almost instantly. Weighing the benefit of a device against its residual risk, however, remains a human judgment shaped by each organization's own risk tolerance, not something AI can determine. He also warns against "AI slop," content that looks polished and authoritative in design file generation but may not hold up to scrutiny. To guard against it, Salvato recommends building skill documents, akin to work instructions, that teach an AI tool a defined methodology, paired with the subject-matter expertise needed to recognize whether the output is sound. As AI takes on drafting work once done by junior engineers, Salvato frames the professional's role as shifting toward reviewing, distilling, and ultimately approving what the tool produces. Evaluating an AI-generated DFMEA or safety risk analysis, he stresses, demands the same depth of training and experience organizations have always needed to sign off on that work. AI can speed up the paperwork of risk management, but it cannot own the judgment behind it. Key Takeaways 0:00 Looking back at Part 1 and introducing today's topic: AI and FMEA 1:45 Reintroducing guest John Salvato, founder of Design Quality Services 4:35 Where AI can responsibly speed up FMEA today 7:45 Why AI cannot weigh benefit against residual risk 10:45 AI slop and the risk of polished but unreliable design file content 15:30 How AI is reshaping the engineer's role from creator to reviewer 20:00 Training quality professionals to evaluate AI-generated output Contact us at [email protected] Contact Dr. John Salvato here: https://www.linkedin.com/in/john-salvato-ph-d-987b1b6/.
Welcome to Automating Quality, the life sciences–focused show exploring how quality, risk, and technology intersect to modernize regulated environments. This is the first installment of a two-part series on design controls and risk management in medical device development with guest Dr. John Salvato, founder of Design Quality Services and an ISO 13485 lead auditor with over 30 years of experience in medical device and automotive engineering. Together with our host Philippe Gaudreau, they argue that as devices grow more connected and AI-enabled, foundational tools like design controls and FMEA matter more, not less. John traces FMEA's origins from the U.S. military in the 1940s to NASA's Apollo program and Ford's automotive use in the 1970s. He contrasts legacy devices like stethoscopes, with their handful of known failure modes, against today's wearable and connected devices, where added functions and data flows multiply the failure modes engineers must account for. The conversation turns to new risks from connectivity and embedded AI/ML models, and to "thin data,". The episode's throughline is that design controls and FMEA, far from outdated paperwork, remain the engineering backbone that novel, complex devices depend on. Key Takeaways 01:19 Introducing today's topic: design controls and FMEA 02:33 Introducing today's guest, Dr. John Salvato 06:27 The origins of FMEA, from military use to NASA and automotive 10:44 Why design fundamentals matter more as devices grow complex 16:49 New risk categories from wearable and connected devices 20:53 The concept of "thin data" in novel device risk estimation 24:14 The QMSR change and the end of legacy device exemptions Contact us at [email protected] Contact Dr John Salvato: https://www.linkedin.com/in/john-salvato-ph-d-987b1b6/
Welcome to Automating Quality, the life sciences–focused show exploring how quality, risk, and technology intersect to modernize regulated environments. In the third and final episode of this series, Philippe closes out the conversation with Niyati Patel, Strategic Quality and Compliance Advisor, returning to explore how life sciences organizations can shift from reactive to proactive quality as AI adoption matures. Niyati distinguishes reactive quality, asking what went wrong after the fact, from proactive quality, asking what pattern signals a problem before it occurs. Drawing on a CDMO case, she shows how AI can connect deviations that look unrelated in isolation, such as operator error, equipment malfunction, and process inadequacy, into an actionable trend. The conversation turns to the industry's move toward integrated architectures connecting ERP, EQMS, and LIMS. Niyati illustrates this with a supplier quality example, then explains how a CSA approach to system interfaces shifts validation focus from exhaustive test scripts toward risk-based confidence in a system's intended use. Returning to AI, Niyati outlines five factors to weigh before moving a tool from experimentation to operational use: business value, risk, data quality, human oversight, and scalability. She walks through validating an AI tool built to accelerate investigation research, applying the same risk-based, CSA-style rigor used for any GxP system. AI can accelerate the pace of quality work, but the judgment behind every decision remains with the quality professional. Key Takeaways 00:22 Introducing today's topic: from reactive to proactive quality 01:20 Reintroducing guest Niyati Patel and catching up since Parts 1 and 2 05:34 What separates reactive quality from proactive quality 06:32 Case example: connecting unrelated deviations into a single trend 10:36 The shift toward enterprise-wide, integrated system architectures 14:55 Case example: integrating ERP and EQMS for supplier quality 18:29 How CSA shifts validation from test scripts to risk-based confidence 25:44 Five factors for moving an AI tool from experimentation to operation 26:46 Case example: using AI to accelerate investigation research 31:17 Why AI supports quality decisions rather than replacing them Contact us at [email protected]. Get in touch with Niyati Patel here: https://www.linkedin.com/in/niyatipatel/.
Welcome to Automating Quality, the life sciences–focused show exploring how quality, risk, and technology intersect to modernize regulated environments. In Part 2 of this three-part series, Mandy and Philippe continue the conversation with Niyati Patel, Strategic Quality and Compliance Advisor, shifting from theory to execution: how organizations are operationalizing AI in GxP environments. This episode dives into the practical realities of AI governance, focusing on lifecycle management, data boundaries, and the human role in processes involving AI. The discussion unpacks how organizations can structure AI frameworks around intended use, risk classification, data governance, and continuous monitoring, highlighting that AI success is driven less by the model and more by people, policies, and control systems. The conversation also explores what can and cannot be shared with AI tools, outlining clear distinctions between acceptable, restricted, and prohibited use cases. From SOP generation to critical quality decisions, Niyati breaks down how leading organizations are defining guardrails to enable safe adoption. Finally, the episode emphasizes that AI is an assistant, not a decision maker. Key Takeaways 01:22 Looking back at part 1 02:00 Introducing today's guest Niyati Patel 05:00 How do organizations safely use AI right now? 07:55 Continuous monitoring is critical for systems that evolve over time 10:10 Which data must be protected when giving access to data to your AI? 11:30 How do you get comfortable using AI as an organization? 15:51 What are some good use cases for AI use in regulated industries? Please contact us at [email protected] if you have questions or comments. Mandy Gervasio Niyati Patel Philippe Gaudreau
Welcome to Automating Quality, the life sciences–focused show exploring how quality, risk, and technology intersect to modernize regulated environments. In this episode, Mandy and Philippe are joined by Niyati Patel, Strategic Quality and Compliance Advisor with 20+ years of experience across pharma, biotech, medtech, and regulated digital implementations. Niyati specializes in the intersection of quality, validation, and digital transformation. Together, they unpack how validation is evolving in the transition from Computer System Validation (CSV) to Computer Software Assurance (CSA). The discussion challenges common misconceptions, emphasizing that the real shift is toward risk-based thinking, lifecycle control, and deeper system understanding rather than documentation-heavy compliance. The conversation then moves into AI. Niyati outlines how organizations should rethink validation frameworks to focus on boundaries, governance, and human oversight, positioning AI as decision support rather than decision-making. Through practical examples and real-world use cases, this episode sets the foundation for understanding how quality leaders can responsibly adopt AI while maintaining compliance, control, and trust. Key Takeaways 01:09: Introducing today's topic 03:24: Niyati's background 05:43: What is CSV (and its limitations) 09:06: What CSA actually changes 10:41: Validation rigor is NOT reduced 14:00 Why AI breaks traditional validation 16:23: Why AI must be used for decision support, not autonomous decision making 19:13: Example of AI validation framework 23:40: What validation now proves Please contact us at [email protected] if you have questions or comments. Mandy Gervasio Niyati Patel Philippe Gaudreau



