SFSA Casteel Reporter – September 2026

Developing Foundry Tools with AI Webinar

Want to learn how to use AI to do something for your foundry?  On September 28^th SFSA members were able to learn about vibe coding from our University of Utah research partner, Taylor Sparks.  The webinar utilized a relevant example for foundries: pointing an AI agent at a pile of disconnected data such as ERP exports, lab results, heat treat chart printouts, customer POs, spec tables and then watching it produce a certified material test report.  You do not need to know how to code – if you know what a good cert looks like, you already have the skill that matters.  The recorded webinar is available to members on the SFSA wiki: https://wiki.sfsa.org/index.php?title=SFSA_Webinars#Vide_Coding_at_the_Foundry

Castings with RT Quality details Needed

SFSA is working with Georgia Southern to develop Phased Array Ultrasonic (PAUT) inspection of castings. To evaluate PAUT, many parts with internal quality details must be inspected and compared to RT results. If you have a casting with quality details, RT results, that weighs less than 200lbs, and can fit into a 36”x36”x36” space; can we inspect it? This is an ongoing request, as a lot more data is needed to move the technology forward. Please keep this in mind and contact jared@sfsa.org if you have parts or questions.

80th T&O Workshop and Conference

The T&O will be December 9-12 in Chicago.  This year will feature a special workshop on sand control Wednesday morning.  The afternoon workshop scheduled topics are metallurgy, burn-in sand and I4.0 NDT.  The conference program will cover the full range of technical and operating topics including Next Gen Manufacturing, molding, melting, foundry engineering, finishing quality, EHS, management, and a range of technical subjects.  The program and registration will be available in early October.

Cast in Steel Competition 2027

The excitement is building for the next Cast in Steel Competition, taking place April 22-24, 2027, at the Huntington Convention Center in Cleveland, Ohio, in conjunction with the AFS Casting Congress. This year, teams will go head-to-head to design and produce a Flanged Mace, showcasing the creativity, precision, and power of steel casting as a modern manufacturing process.

The 2027 competition will also be documented as a TV series and is expected to appear on a major streaming platform. This is an exciting opportunity to showcase the talent, innovation, and teamwork behind Cast in Steel and the steel casting industry.

Missed an episode or just want to relive some of the exciting moments? You can now catch up on the entire first season whenever it’s convenient for you.

All six episodes of Cast in Steel are now available to watch anytime onYouTube!

Market News

US Economy and Manufacturing Update

Source: ITR Economics

The U.S. economy remains on a moderate growth path, with ITR reporting real GDP growth of approximately 2.1% and continued support from consumer spending and business investment. However, the outlook is becoming more constrained as inflation pressures purchasing power, input costs remain elevated, and higher borrowing costs weigh on interest-sensitive activity. ITR expects economic momentum to moderate moving into 2027 rather than develop into a broad downturn, with growth increasingly concentrated in areas such as AI/data-center investment, power infrastructure and defense.

U.S. manufacturing and industrial production continue to expand in the second half of 2026, but ITR expects the cycle to reach an inflection point and flatten considerably in 2027. Capital-goods demand remains healthy but is increasingly concentrated in defense, AI/data centers, power and electrical infrastructure, while more traditional industrial markets are expected to experience slower growth or modest declines. Heavy-duty trucks, metalworking machinery and defense received forecast upgrades in the September outlook, while oil and gas is expected to be essentially flat next year. The result is a distinctly K-shaped manufacturing outlook: opportunities remain strong in selected high-investment markets, but manufacturers should not assume that 2026 growth rates will carry into 2027 and should plan around market-specific demand rather than a broad industrial expansion.

SFSA Business Trends

To benchmark your facility with other steel foundry members, SFSA encourages you to participate in the monthly SFSA business trends survey – only participants have access to the results. Please complete the business trends survey for August 2026: https://www.surveymonkey.com/r/SFSA_BTB_Aug26. Survey results are provided to participants the following month.

Navy-Sponsored Workforce Program Available to SFSA Members

The casting industry plays a critical role in America’s Defense Industrial Base. Meeting the growing needs of the U.S. Navy and our nation’s defense manufacturers requires more than production capacity. It requires companies that can find the right people, develop them, and keep them.

SFSA members are welcomed to participate in the Talent Pipeline Program (TPP) through the NFFS cohort of foundries.

Sponsored by the U.S. Navy, TPP is available at no cost to participating companies. It is the first program of its kind dedicated to helping manufacturers address one of industry’s greatest challenges: acquiring the right talent and retaining them.

Built around the philosophy Hire for Fit. Train for Skill. Retain for Life., TPP provides practical tools and hands-on coaching to help companies:

• Attract and hire employees who are the right fit
• Improve onboarding and the critical first 90 days
• Develop stronger frontline supervisors
• Reduce turnover and improve retention
• Measure workforce performance and results

The program is already producing results. At TPi Arcade, implementation of TPP’s Leadership Retention Training and 5th Metric Scorecard helped one department improve efficiency from 85% to 110% within months.

SFSA participants will join the NFFS TPP cohort, learning alongside other foundries and receiving direct support from trained NFFS Network Coaches. Interested in participating in this free, Navy-sponsored program? Contact Bill Padnos, NFFS Director of Workforce Development, at bill@nffs.org to learn more or visit https://www.nffs.org/tpp.

Casteel Commentary

This month’s commentary is on the rapidly changing role of Artificial Intelligence (AI) within the foundry industry.  A year ago, AI seemed to have a lot of hype but not a lot of helpfulness when it came to casting.  When used in a closed environment with SFSA resources, the results were initially lacking.  A second iteration improved the results but while they were reasonably factual, they lacked in being useful – best described as something written with the grammar of a 20-year-old but the metallurgical mind of a 2-year-old.  In fairness, AI is just software and if it has a limited dataset, then just like you or me, it simply doesn’t know.  This was described as a big “A” and a little “i”.  In other words, it seemed more artificial than intelligent.

Over the past half year, an agentic approach to both leverage other sources and to validate responses improved our results.  The role of AI in most steel foundries was still very limited.  Within the last three months, this has rapidly started to change.  The AI assistants have become more capable – now better able to use unstructured data.  The opportunities for a foundry have become clearer – use AI to create code to reduce administrative work, build dashboards for data output, analysis of heat data, etc.  AI is not actively used but instead creates a tool (code) that runs and provides service.  This resolves two primary concerns of AI – posting to a public/cloud AI resulting in your data not being just yours anymore, and if AI is directly accessing your data, does it become corrupted. 

This evolution of AI for foundries is currently defined by SFSA’s AI working group with five levels:

1. Replace Google search and maybe some small individual uses such as e-mail or spreadsheet help.
2. Use AI to Vibe Code apps and programs but still have hard code, not AI run but may use some small AI analysis.
3. True AI to analyze data and make decisions/recommendations; advanced engineering and processes begin to be AI controlled and assisted.
4. Physical AI with humanoid or traditional robots programmed and controlled by AI with ~20% production by robots.
5. AI-Foundry with all critical operations and decisions done through AI with AI robots and humanoids.

Getting started is as simple as “talking” to an AI assistant (Claude being the most popular but Copilot is typically accessible to most through Microsoft products).  Asking AI to simply create or do something will likely not result in useful outcome.  AI like any communication is best done iteratively or refined in back-and-forth discussion.  Here are some tips on how to best work with AI on a project:

• Give it context and a role. “I’m a quality engineer at a steel foundry writing a cert for an oil and gas customer”
• Give it the source material (the spec, an old cert, the test data) and tell it to work only from that.
• Describe what good output looks like: format, length, audience, and an example if you have one.
• Ask for honesty about uncertainty: “If you don’t know, say so. Don’t guess.”
• Ask for its reasoning and sources: “Show your work and tell me where each number came from.” Then spot check a couple.
• Before it starts, ask “What else do you need from me to do this well?”
• Treat the first answer as a draft and tell it what to fix.

We must also consider that AI is not just a tool for us but is now being used in a similar manner by customers.  Websites became our billboard or Yellow Pages listing; however, if websites are not structured properly, they now may be overlooked by AI.  Recommended for having an AI agent ready website:

• MCP (Model Context Protocol)
• Web MCP (Web Platforms Protocol)
• NL Web (Natural Language Web)
• ACP (Agentic Commerce Protocol)
• UCP (Universal Commerce Protocol)
• AP2 (Agent Payments Protocol)

AI (or search engines) cannot read embedded apps.  Therefore, search functions such as a directory or alloys need to be published as plain, crawlable pages or a simple structured data feed.  Enabling a natural language search instead of guessing keywords means a user can ask, “what cast alloy should I use for low temperature impact service?” and get a direct answer with links to the references for additional information.

SFSA research is looking to deploy a range of AI solutions to advance the steel casting industry.  AI is a resource as natural language assistant (Development Of A Standalone Retrieval-Augmented Generation System for Steel Casting Wiki).  But, AI is also integral to new Foundry 4.0 technology:  Internet of Things (IoT), Machine Learning (ML), Operational Technology (OT) and vision technology.  Our primary focus has been on assisting the NDT inspector to improve the repeatability and reproducibility:

• Quality Detail Automated Recording and Tracking
• Automatic Non-Destructive Testing of Steel Castings Using AI and XR
• Automated Inspection Assistance System with MPI
• Digital Surface Characterization
• Radiography Automated Image Analysis

As our industry continues to pursue AI applications, it is important to have and routinely review a company AI policy.  Some common practices:

• Restrict or know who is using AI tools and access to information used follows existing data/file rights limitations
• Provide approved AI tools and training for their use (including safe and ethical)
• Generally open/public AI are not permitted or restructured
• Information such as copyright material, personal identification and company IP are not allowed to be used
• Review restrictions based on agreements with customers
• Critical decisions still require a Human In The Loop (HITL)

It is also important to have and improve company cybersecurity; especially as we shift from digital AI to physical AI solutions.  A few helpful resources:

• Global framework for securing automation and control – ISA/IEC 62443 Standard–systems
• The foundational 5-step framework for securing industrial environments – SANS ICS Critical Controls
• Definitive guide to securing industrial control systems – NIST SP 800-82
• Foundational guide to Zero Trust Architecture – NIST SP 800-207
• AI Risk Management Framework (RMF) – NIST AI Resource Center (AIRC)

On a different topic than AI, I would briefly like to express how very grateful I am to have heard from so many people and to know I have such great support as I take on a new role for the organization; this was truly quite humbling.  As I reflected on my career, I know I have learned so much from so many throughout my years in the industry – thank you all!  This willingness to help each other and work together has always been key to Steel Founders.  The fact that our industry gets together to share technical and operating papers or hosts foundry tours is unique and demonstrates our core values of what this industry means to us.  But we also work broadly with suppliers, customers, researchers and government partners, and within standard communities or other organizations to both help and to have the support of others to help the industry holistically.  In the years to come, I will continue to pay it forward by helping others in our community.  My goal is to foster a community for the founders of steel and those who support them with a future that returns dividends to all who have invested their time helping.  I appreciate everyone’s support as we work to advance the steel casting industry.  My icing on the cake was when my daughter told me that she was proud of me after I let her know the news.

David Poweleit