The only thing that's gonna matter over the next four or five years as we're in a critical crunch of capacity is your relationships. Who do you know? Who can you call? People are gonna be the only thing that move the needle here. If you don't have a network, you better get moving on it. β @MatthewChang, EP 014
America is heading into a critical capacity crunch across every layer of manufacturing. Trucks, electrical gear, generators, turbines, and engineering firms. The bottleneck is not capital or policy anymore, it's relationships and lead times. The companies that start building their networks now are the ones that will still be moving in 2026.
There's so much conversation out there about VCs who want American dynamism and want to invest in America, who have only been putting money behind AI slop or defense companies. What SendCutSend have done is paving the way for venture and private capital to adjust and open their eyes to what is possible in this new era of manufacturing. β @ThaaatColin, EP 014
Ford is repurposing its Kentucky battery facility to produce battery energy storage systems for utilities and data centers. Most factories in America are running on unreliable power right now, paying full price for electricity they can only use 60% of. Battery backup is not optional anymore, but it's basic infrastructure.
@ChangRobotics has been deploying AI vision stacks for over a year and the results are clear.
One visual intelligence layer informs every robot system in the cell and eliminates the need for most of the hardware underneath it. Faster deployment, less complexity, and clients who love the results.
This is what factory design looks like right now.
ποΈ @MatthewChang on the ReForge Podcast
π Full episode link in the comments
Start as far away from your core offer as possible. That's sales and marketing. As you get closer to your core offer, get lots of context and tone down the percentage of AI in the final work. By the time you're on the last few passes, you have a human in there. You're not trusting AI for anything on those final passes. β @briankeithai, EP 013
Brian Keith's minimum standard for any AI output is two layers of human contact. A transcript of real people talking, voice notes from the actual subject matter expert, writing based on work that person has already done. The moment you remove the human from the loop, AI starts passing bad information to itself and calling it true.
You can trust the LLMs want what's best for them, which is not what's best for you. They have their own agenda and it's not to serve you the small business person, it's not to serve you the factory owner. Their problem is how to go maximize their revenue. β @briankeithai, EP 013
Machine learning and LLMs are not the same thing and most people talking about AI in manufacturing don't know the difference. Machine learning is purpose built for structured data, sensors, and precise outputs. LLMs are great at unstructured conversation but hallucinate constantly. The opportunity is in bridging the two.
Brian's minimum standard for any AI output is two layers of human contact. A transcript of real people talking, voice notes from the actual subject matter expert, and articles written from before. The AI writes based on all of that, and a human reviews the final pass.
The moment you remove the human from the loop, AI starts passing fake information to itself and calling it true.
ποΈ @briankeithai, CMO of Chang Robotics, and @ThaaatColin on the ReForge Podcast
π Full episode link in the comments