Daikin Deploys AI to Unlock Decades of Trapped Engineering Designs

SAN FRANCISCO, CA – March 19, 2026 – In a significant move to bridge the gap between industrial legacy and artificial intelligence, global manufacturing leader Daikin Industries has fully deployed a new AI platform from San Francisco startup Outerport. The partnership aims to solve one of the manufacturing sector's most persistent challenges: unlocking decades of valuable engineering knowledge trapped in unstructured documents, and in doing so, pave the way for a new era of AI-driven design and innovation.

Daikin, a world leader in heating, ventilation, and air conditioning (HVAC), will use Outerport's platform to parse and extract structured, machine-readable data from its vast archives of complex engineering drawings, diagrams, and technical documents. This initiative is designed to be the foundational step in building a fleet of specialized AI agents capable of automating and accelerating core engineering workflows.

The Unstructured Data Bottleneck

While software development has been transformed by AI code assistants and autonomous agents, the manufacturing industry has largely remained on the sidelines, constrained by the nature of its data. For decades, the most critical intellectual property of industrial companies—the technical know-how, design specifications, and operational knowledge—has been stored in formats like paper scans, PDFs, and 2D drawings. This visual, unstructured data is illegible to standard AI models, creating a bottleneck that has stymied automation in core design processes.

This challenge goes far beyond simple text recognition. Engineering drawings are a complex language of their own, composed of intricate diagrams, specific symbols, geometric dimensioning and tolerancing (GD&T) annotations, and handwritten notes. Traditional Optical Character Recognition (OCR) tools often fail to capture this contextual and visual complexity with the accuracy required for mission-critical engineering tasks. Without a way to reliably translate this visual language into the structured formats that AI understands, manufacturers have been unable to leverage their most valuable asset: their own historical data.

A New Breed of AI for the Factory Floor

Enter Outerport, a startup founded by former research engineers who led AI development at tech giants NVIDIA and Meta Platforms. The company was created specifically to solve this industrial data problem. By combining proprietary computer vision technology with advanced multi-modal large-language models (MLLMs), Outerport's platform learns to read and interpret engineering documents like a human expert.

Instead of just extracting text, the system converts complex visual information into AI-ready structured formats like JSON. This high-fidelity extraction allows AI agents to process, analyze, and reason over a company's entire engineering history. This migration of top-tier AI talent from consumer tech to solve fundamental industrial problems signals a growing trend of applying cutting-edge research to reshape traditional sectors.

Daikin selected Outerport after internal trials highlighted the immense difficulty of achieving the necessary precision. "Converting unstructured data into structured data has been a long-standing challenge and a major area of technical interest for us," said Shohei Hido, Executive Engineer at the Technology Innovation Center, Daikin Industries. "Upon evaluating Outerport's technology, we were able to achieve a dramatic leap in parsing accuracy compared to conventional approaches. Beyond their technical capability, we highly value their agile development process and ability to customize the platform to our specific needs."

Daikin's Strategic AI-Powered Future

This partnership is not an isolated experiment for Daikin; it is a cornerstone of a meticulously planned and well-funded corporate strategy. The company's long-term strategic plan, "FUSION 25," explicitly identifies digitalization as a key driver of transformation, backed by a committed investment of 180 billion yen (approximately $1.2 billion USD) into digital technologies. Daikin has been repeatedly recognized by the Japanese government as a "Digital Transformation (DX) Brand" for its enterprise-wide efforts to integrate AI and IoT.

The Outerport deployment complements Daikin's other AI initiatives, such as using AI-powered wearable devices to create a video library for training service engineers and developing collaborative platforms that use air conditioner data to improve office environments. By making its engineering data accessible, Daikin is laying the groundwork to scale its AI capabilities from operational support to the core design process itself, aiming to enhance efficiency, reduce downstream errors, and accelerate product development cycles.

Navigating Security in a Competitive AI Market

Deploying AI to handle a company's "crown jewel" intellectual property introduces significant security risks. The centralization of sensitive design data for AI training creates a high-value target, making data security and IP protection paramount. Daikin's choice of Outerport was explicitly tied to the platform's ability to meet the stringent information security standards required by high-security manufacturers, a crucial factor for any company entrusting its core IP to a third-party system.

The market for industrial AI is booming, with analysts projecting a compound annual growth rate (CAGR) of over 35% for AI in manufacturing, potentially reaching over $100 billion by 2033. While major industrial software vendors like Siemens and PTC are integrating AI into their platforms, Outerport's specialized focus on high-fidelity data extraction for AI agents carves out a critical niche. The company’s mission extends beyond data conversion to enabling an "AI-driven engineering system" capable of performing automated design checks and analysis.

By successfully transforming its historical data into an active asset, Daikin is not just optimizing its current operations. It is building a foundation for a future where AI agents can collaborate with human engineers, test designs virtually, and catch potential issues early in the process, setting a powerful precedent for the entire manufacturing industry.