When the Expert Retires, Does the Answer Leave with Them? with Brian Kaufman

Brian Kaufman is co-founder and chief product officer at Strudel, an AI startup that gives technical support teams real-time context when issues hit and detects error patterns before they become customer-facing problems. His background is in software product development and healthcare technology, but the problems he solves map directly onto what manufacturing companies face every day: institutional knowledge that lives in people's heads instead of in systems, and support teams solving the same problems over and over from scratch.

Brian and Lisa dig into what it actually costs when undocumented knowledge walks out the door, how AI can serve as a partner to experienced technicians without replacing their judgment, and what a realistic starting point looks like for manufacturers considering AI-assisted support.

Key Takeaways for Manufacturing Leaders

The information exists. It just isn't findable. Most organizations aren't starting from zero. The knowledge is in old tickets, emails, PDFs, and people's heads. The first step is figuring out where it lives, not waiting until it's gone to wish you'd captured it.

Start before someone gives notice. The best time to document institutional knowledge is before retirement is on the horizon. If the person with 30 years of experience is leaving in two months, you're already behind. Build the habit now.

Ask the question nobody's asked. One of the most powerful knowledge-capture prompts: "What do you know that nobody's ever asked you about?" Experienced technicians have built sensory libraries over decades, knowing what a machine sounds like, feels like, and smells like when something's about to go wrong. That knowledge doesn't survive a standard exit interview.

AI should be an investigator, not an oracle. Strudel presents troubleshooting suggestions as hypotheses grounded in real historic tickets, not confident declarations. When a technician is standing at a machine, that distinction matters. Build trust by showing your work, not by pretending to be certain.

Machine context cuts investigation time dramatically. Knowing the model, firmware version, and configuration immediately narrows the problem space. A known defect affecting machines shipped between two specific dates becomes a one-step fix rather than a multi-hour troubleshooting process.

Don't use LLMs for everything. Strudel uses traditional statistical analysis and machine learning for the heavy lifting, reserving large language models for last-mile summarization. The result is fewer hallucinations and answers that are easier to monitor and evaluate for quality.

Your most skeptical expert is your most important early win. Find the team member with deep expertise who doubts new tools and figure out what task they hate doing most. Solve that one problem well, and you've turned your biggest skeptic into your biggest advocate.

Start small, find wins, then expand. Don't try to solve everything at once. Pick one product line, one process, one team. Get good at that before scaling. Build in an early checkpoint to evaluate and adjust.

Involve your people before you buy anything. Coming back from a trade show with a shiny new system your team had no input on is a reliable way to end up with a system nobody uses. Ask what's causing the most pain. Start there.

Working top of license matters on the shop floor too. When your most experienced technician is answering the same question for the fifth time, that's a waste. Free those people up for novel problems, mentoring, and process improvement.

About Strudel

Strudel embeds directly into existing support ticketing systems and generates troubleshooting steps within 15 to 20 seconds of a ticket coming in, drawing on historical tickets and documentation specific to the machine, software version, and configuration in question. As more tickets are resolved, that knowledge becomes part of the system, making new staff onboarding faster and institutional knowledge more durable.

Connect with Brian Kaufman

Email: brian@strudel.io Website: www.strudel.io

The Manufacturers Network Podcast features long-form conversations with leaders, practitioners, and experts across manufacturing, the skilled trades, and industrial sectors. Host Lisa Ryan, CSP, MBA, is a keynote speaker, author of 13 books, and founder of Grategy.