From shop floor secrets to AI savvy: how expert knowledge is powering the next wave of smart factories

By early 2026, it became clear that the primary bottleneck in American manufacturing productivity was not machinery but the scarcity of tacit expert knowledge residing in individual workers’ heads. Companies like CloudNC demonstrated that embedding this domain-specific expertise into AI systems—such as their machinist-like CNC programming AI—could transform expert judgment into scalable software, accelerating workflows and enabling consistent best practices across hundreds of factories. This shift from manual, siloed tribal knowledge to AI-encoded expertise promises to unlock significant efficiency gains by automating repetitive expert tasks and empowering less experienced workers with strong starting points.

The challenge of capturing tacit knowledge is intensified by an aging skilled workforce and acute labor shortages, with trades like electricians commanding salaries surpassing those of Silicon Valley software engineers, underscoring the urgency of knowledge transfer. Analysts highlight a cultural disconnect in manufacturing where design and production are often siloed, leading to atrophied design-for-manufacturing expertise—a gap that countries like China mitigate through close collaboration between designers and manufacturers. This embedded, context-rich tribal knowledge is notoriously difficult to document but essential for efficient production and must be integrated into AI models to sustain scalable manufacturing.

Squint’s CEO Devin Bhushan emphasizes that tribal knowledge—those undocumented, company-specific insights—is critical for AI to effectively enhance industrial lines. Their fine-tuned 2 billion parameter AI model, which integrates tribal knowledge with comprehensive data sources such as work orders, asset histories, SOPs, and regulatory requirements, achieved a remarkable 78% accuracy on complex manufacturing questions, outperforming generalist AI models like Claude Code and OpenAI File Search by large margins. This demonstrates that encoding expert knowledge into specialized AI systems is not just beneficial but necessary to overcome the core knowledge bottlenecks limiting manufacturing productivity.

Beyond programming, AI’s potential to capture and encode expert knowledge extends across the entire manufacturing ecosystem—including quoting, process planning, scheduling, setup, and inspection—areas traditionally dependent on tribal knowledge. This comprehensive AI integration is pivotal for U.S. industrial policy goals such as reshoring and defense readiness, enabling factories to handle greater complexity and boost productivity with the skilled labor they already possess. As tariffs and subsidies alone prove insufficient, AI-driven knowledge capture emerges as the linchpin for revitalizing American manufacturing competitiveness.

By mid-2026, AI adoption in U.S. manufacturing had surged from exploratory phases to active pilot programs targeting predictive maintenance, quality control, and supply chain resilience, with 92% of companies engaged in AI initiatives according to a Sikich survey. However, manufacturers grappled with fragmented legacy data, a shortage of data science talent, and integration hurdles that complicated scaling AI beyond isolated use cases, underscoring the tension between ambitious digital transformation goals and operational realities.

GFT Technologies exemplified the next wave of smart factory innovation by moving AI from mere defect detection to autonomous physical intervention on automotive assembly lines. Brandon Speweik detailed how their integrated system combines machine vision, robotics, edge computing, and cloud-based root-cause analysis to identify, reposition, or remove defective components in real time, overcoming synchronization challenges that threatened to disrupt high-speed production. This evolution toward execution orchestration systems tightly linking detection, intervention, evidence capture, and continuous learning marks a pivotal shift in manufacturing AI capabilities.

Despite technological advances, trust remains a critical barrier to widespread adoption of autonomous AI decision-making on factory floors. Speweik emphasized that successful AI systems will earn confidence incrementally by delivering measurable results in targeted, high-value applications rather than replacing human judgment outright, highlighting the importance of human-machine collaboration in early deployments.

Squint’s AI system, fine-tuned specifically for manufacturing, demonstrated superior performance over generalist AI models from OpenAI and Anthropic by achieving 78% accuracy on complex industrial queries. CEO Devin Bhushan highlighted the AI’s unique ability to integrate 'tribal knowledge'—the tacit expertise of experienced workers—with formal documentation and regulatory data, creating a rich context layer that augments rather than replaces human operators. This approach addresses the critical challenge of preserving and leveraging experiential knowledge that is otherwise difficult to codify, thereby enhancing frontline modernization efforts.

The integration of legacy industrial equipment with modern AI and cloud systems in automotive production lines has significantly increased complexity and cybersecurity vulnerabilities, as highlighted by Bodo Philipp of MHP Consulting UK. OEMs now face systemic risks amplified by an expanded supplier attack surface spanning hundreds of Tier-1 to Tier-3 suppliers, with legal accountability mandated under UNECE R155 regulations. The Jaguar Land Rover cyberattack, which disrupted 5,000 suppliers and delayed recovery until early 2026, underscores the critical need for adherence to standards like ISO/SAE 21434:2021 to manage cybersecurity risks across the vehicle lifecycle and supply chain.

Automotive manufacturers are rapidly advancing from AI-powered defect detection to autonomous physical interventions on the factory floor, as demonstrated by GFT Technologies’ system that integrates machine vision, robotics, and cloud AI for real-time root-cause analysis and corrective actions. Brandon Speweik emphasizes that this leap requires seamless orchestration of computer vision, robotics, operational data, and human escalation workflows, all synchronized to meet the stringent timing and precision demands of high-speed automotive lines. Despite these technological strides, building trust through measurable results in high-value applications remains essential for broader adoption of autonomous AI decision-making.

The industrial automation market has evolved into a strategic imperative for automotive OEMs, driven by the demands of EV and software-defined vehicle architectures. Investments in smart factories now leverage AI-driven analytics, robotics, and predictive maintenance to address labor shortages, reduce operational risks, and manage production complexity. This shift from isolated automation to connected digital ecosystems—featuring integrated robotics, sensors, digital twins, and cloud platforms—enables real-time decision-making and has positioned China as a global leader due to its expansive EV manufacturing and smart factory deployments.

AI and machine learning are delivering tangible performance gains in automotive manufacturing, with reported reductions of up to 50% in unplanned downtime and throughput improvements of 5% to 7%. The industry is moving beyond traditional automation in body and paint shops into complex domains like electronics assembly and production logistics, using AI to manage growing manufacturing complexity and enhance decision-making. However, as James Glasson of Rockwell Automation notes, the critical challenge lies in scaling these AI-automation capabilities effectively, with disparities in adoption creating widening gaps in quality, uptime, and supplier competitiveness.

By mid-2026, the automotive industry is decisively shifting from hardware-centric vehicles to software-defined vehicles (SDVs), with projections indicating that by 2035 nearly 80% of new cars will be AI-powered and built on centralized computing architectures. This transformation hinges on modularity, user experience, and foundational trust in cybersecurity, moving OEMs away from fragmented hardware supplier models toward integrated technology partnerships that synchronize software, AI, and UX platforms. As Roshan Kulati, CEO of SDVerse, highlights, open source and AI advancements are accelerating modular vehicle architectures that decouple software from hardware, although automotive software will remain tightly controlled due to stringent safety and reliability requirements.

The rise of AI-native engineering is dramatically compressing vehicle development cycles and creating stark performance divides between legacy OEMs and digitally native newcomers. Secondmind’s Engineering AI, now integrated into SCAE’s services for Japanese OEMs, cuts simulation needs by up to 80% and calibration time by over 50%, enabling companies like Mazda to slash engineering hours and accelerate innovation. Yet, entrenched traditional engineering processes—largely unchanged for two decades—pose significant adaptation challenges, underscoring the widening gap as digital-native firms leverage AI to speed workflows within a year, outpacing slower, established automakers.

The software-defined vehicle era introduces complex cybersecurity and regulatory challenges, especially on the manufacturing floor where legacy machinery—often running proprietary or decades-old software—was never designed to withstand modern cyber threats. Bodo Philipp emphasizes that under UNECE R155, OEMs bear legal responsibility for cybersecurity across the entire vehicle lifecycle and supply chain, necessitating rigorous governance frameworks like ISO/SAE 21434:2021 to manage risks and supplier integration. Given the impracticality of frequent legacy system updates due to costly downtime, structured vulnerability management and compensating controls are critical to balancing security with uninterrupted production.

As automotive electronic architectures evolve from distributed ECUs to centralized and zonal designs, the industry gains enhanced agility for faster feature deployment and improved cybersecurity, with AI and digital twins playing pivotal roles. However, regional nuances such as India’s complex road environments demand localized ADAS data to ensure software efficacy, highlighting the need for OEMs to strategically retain core vehicle intelligence development in-house—covering system architecture, software platforms, and cybersecurity—while outsourcing components like semiconductors and sensors. This balance is essential for maintaining competitive advantage in the global SDV landscape.

By mid-2026, China solidified its position as a global powerhouse in automotive automation, leveraging its vast EV manufacturing ecosystem and extensive smart factory deployments to drive innovation. Chinese automakers are not only embedding AI into infotainment but are advancing towards AI-defined vehicles that personalize driving experiences through sophisticated learning of individual styles and battery management. This domestic focus on competing within a crowded local market, combined with strategic export initiatives that boast higher profitability and capacity utilization, underscores China's deliberate and evolving role in the global competitive landscape.

India’s automotive sector is undergoing a pivotal transformation from hardware-centric to software-driven innovation, emphasizing the development of domestic IP, product architectures, and software capabilities. Industry leaders like Mahindra’s Dr. Shankar Venugopal and Ather’s Swapnil Jain highlight the critical need for localized real-world ADAS data tailored to India’s complex traffic conditions, while advocating for OEMs to own core vehicle intelligence technologies in-house and collaborate strategically with specialized partners. This ecosystem collaboration is further supported by a shift towards centralized and zonal electronic architectures, enabling faster feature deployment and enhanced cybersecurity.

Across North America and globally, the automotive industry is witnessing a strategic pivot where software-defined business models are reshaping competitive dynamics. Automation vendors now compete primarily through advanced manufacturing execution systems, cloud-connected platforms, digital twins, and analytics rather than hardware alone, reflecting the growing importance of integrated ecosystem partnerships. Additionally, OEMs and suppliers are expanding the scope of industrial robotics beyond traditional tasks into complex areas like battery assembly and collaborative manufacturing, highlighting the necessity of seamless collaboration within the automotive value chain to address evolving production complexities.

Chinese automakers’ export strategies reveal a learning curve shaped by early challenges in Europe, prompting refined approaches in diverse markets such as South America, Central America, and Northern Africa. This adaptive global expansion, combined with a robust domestic AI-driven vehicle development, illustrates how China balances intense internal competition with deliberate international market penetration, setting a competitive benchmark for other regions navigating their own ecosystem collaborations and technological advancements.

Legacy machinery in automotive manufacturing continues to pose significant cybersecurity challenges due to its heterogeneous and long-lived nature, often running proprietary systems that cannot be updated without disrupting production. As Bodo Philipp, CEO of MHP Consulting UK, highlights, ISO/SAE 21434:2021 provides a crucial framework for managing these risks by standardizing cybersecurity processes across vehicle development and production lifecycles, enabling OEMs to meet their regulatory obligations under UNECE R155 to oversee the entire supply chain's security.

The devastating Jaguar Land Rover cyberattack, which disrupted thousands of suppliers and delayed sector recovery until early 2026, underscores the critical need for comprehensive supply chain governance and cross-industry collaboration to build trust and resilience. As the automotive industry accelerates toward software-defined vehicles (SDVs), projected to constitute up to 80% of new sales by 2035, cybersecurity and data trust emerge as foundational pillars of competitive advantage rather than mere compliance exercises.

Scaling AI integration on the factory floor demands moving beyond defect detection to autonomous physical intervention, a complex orchestration of computer vision, robotics, edge computing, and cloud-based root-cause analysis. Brandon Speweik of GFT Technologies emphasizes that trust in AI autonomy remains a critical barrier, with successful implementations earning confidence gradually through targeted, high-value applications rather than wholesale replacement of human judgment.

The shift toward modular vehicle architectures, championed by OEMs and supported by open source initiatives and AI advancements, promises to decouple software from hardware for greater flexibility. However, as Roshan Kulati, CEO of SDVerse, cautions, automotive software will not become fully plug-and-play or downloadable like consumer apps anytime soon due to stringent requirements for reliability, safety, and traceability, necessitating continued innovation in governance and regulatory clarity, especially with proactive U.S. government engagement to accelerate safe AI deployment.