Technology

A path with no footprints and no 'right answer' is the one most worth forging.

Advancements in technology, including AI, have finally brought within reach challenges that were once impossible to solve.
To take on a universally shared problem of extraordinary difficulty, one that has remained unsolved for over a century.
Here is a challenge where you can fully bring your professionalism to bear.

CTO Message

CTO

Aki Kobashi

Since the Industrial Revolution, humanity has advanced from steel to steam to silicon – from materials to machines to digital. Yet, manufacturing remains the backbone of our lives today.

Every product we touch is built by the dedication and skill of people across the supply chain.In today's dynamic world, our creations are becoming more complex and technologies are evolving faster, all while consumer expectations for quality and variety continue to rise. At the same time, our most experienced veterans are retiring, and the next generation must master an ever-expanding body of knowledge.

However, to exceed the limits of what an individual can manage, we simply cannot "digitalize" our way out. Things must still be made, transported, and delivered. Mistakes in the real world cannot be undone with a few keypresses.We must leverage the digital world and connect it to the physical world to take manufacturing to the next level.

By combining cutting-edge technology with world-class manufacturing, we aim to amplify the human capability in order to build the next generation of manufacturing. Join us, and help build the future.

Confronting unknown challenges with the full force of technology

To solve the complex challenges of manufacturing, we orchestrate a diverse set of technologies.
These technologies do not operate in isolation. They interact and reinforce one another to bring a global product to life.

Structural Challenges in Manufacturing

N:N:N data complexity and the barrier of unstructured information

Latent Complexity

In supply chains where a single product involves thousands of components and numerous companies, critical information is buried as unstructured data.

Barrier to Potential

The sheer volume and variety of data—far exceeding other domains—creates a level of complexity that no single algorithm can fully optimize, standing in the way of effective data utilization.

Irreversibility of the physical world & the challenge of building an “OS”

Latent Complexity

In the physical world, there’s no “Ctrl-Z”—failures translate directly into real costs and delays, demanding extremely high reliability from software.

Barrier to Potential

Our goal is to build a “Manufacturing OS” that supports the industry across the board.
This goes beyond delivering individual applications. To create products that work across countless use cases, we need a high level of abstraction grounded in a deep, high-resolution understanding of the field.

Balancing conflicting stakeholder values

Latent Complexity

Siloed information, non-uniform manufacturing processes, and regulations across a global industry—because processes aren’t standardized, each site operates differently, with strong site-specific optimizations in place.

Barrier to Potential

Transformation cannot take hold unless both management rationality and on-site consensus are achieved. One without the other cannot move the industry. A product that bridges both is a prerequisite for real-world implementation.

Because It's difficult. That's why it's interesting

The imperative of speed in achieving non-linear growth

Latent Complexity

Transforming manufacturing is inherently a long-term, cumulative effort that typically unfolds over years. At the same time, a startup must move fast, diving deep into the industry while achieving non-linear growth.

Barrier to Potential

Before incumbents fully mobilize, we need to create global value and set the standard. Operating on this timeline where depth and speed must coexist is both the challenge and what makes it compelling.

CADDi’s Technological Advantage

Structuring Complex Data

By combining machine learning, natural language processing, and knowledge graphs, we extract insights from unstructured data and transform tacit knowledge into reusable forms. Tackling complexities that no single technology can solve, our integrated approach elevates this information into a foundation that supports decision-making on the ground.

Product Evolution Driven by Deep Domain Knowledge

Because we have deep domain knowledge gained from working directly within supply chains, we can understand the context of complex data and structure it effectively. The more the data is utilized, the better we interpret it, allowing the product to continuously evolve.

Development Rooted in the Field

We immerse ourselves in real-world operations and face our customers’ challenges head-on. Designs built with a deep understanding of users and their problems are what create truly valuable products. To achieve meaningful impact, we embrace whatever processes are necessary.

Tech Stack

Frontend

TypeScript
React
Next.js
Apollo Client
Storybook
Vitest, Jest

OpenAPI
GraphQL

BFF

TypeScript (NestJS)
Apollo Server

May not be implemented depending on system requirements.

OpenAPI
gRPC

Backend

TypeScript (Express, Fastify, NestJS)
Rust (axum)
Go, Java (Quarkus)

Other languages may be used depending on specific processing needs.

Algorithm Machine Learning

Python
OpenCV
Vertex AI
Unsloth
Rust
PyTorch
LangChain

Infrastructure

Google Cloud, Kubernetes, Istio, Cloud Run, Cloudflare, Argo Workflows, etc …

Async integration

Cloud Pub/Sub, Cloud Tasks

DevOps

GitHub Actions, Terraform, Conftest Argo CD, Kustomize, Helm, Kyverno Datadog, MixPanel, etc …

Data

Cloud SQL(PostgreSQL), AlloyDB, BigQuery, dbt, Trino, Iceberg, Elasticsearch

Authentication

Auth0, Firebase Authentication

Dev Tools

GitHub Enterprise, GitHub Copilot, Cline, Claude Code, Devin, Figma, Storybook, etc …

Communication

Slack, Confluence, Jira, Miro, Asana

Key Points in Technology Selection

To fully leverage type systems in development, we prioritize technology choices centered around statically typed languages.

While selecting technologies based on product characteristics, we also work on cross-team standardization to achieve overall optimization.

Whole Product

CADDi is not just a software company. We exist to transform manufacturing.
Software, operations, and business all come together to create the experience we deliver. That whole is our product.
We call this mindset “Whole Product” — continuously thinking through and executing how to maximize the value we deliver as a company.

The future of CADDi Technology

Executive Officer, VP of Growth Enablement

Jumpei Nishina

Bringing technology and business together to reshape the world at a defining moment in history.

No matter how technically sophisticated a product may be, it cannot endure if the underlying business structure is weak.
We have been continuously reflecting on a central question: what does it truly mean for both technology and business to succeed together?

The data we deal with in the manufacturing industry is far less structured than one might imagine. For example, even something as simple as part numbers may follow different conventions between specification documents and bills of materials, making it difficult to link documents together. Variations in formats and inconsistencies in notation are effectively endless, which makes it challenging to achieve a level of data structuring and utilization that is practical by simply applying general-purpose AI as-is.

In addition, manufacturing relies heavily on the experience and tacit knowledge of seasoned professionals. Simply collecting visible data is often not enough to explain why a particular decision was made.

Within the vast landscape of manufacturing, there are countless leverage points where technology can meaningfully connect to business outcomes. At this historical turning point defined by the rise of AI, we are committed to defining a new form of software—and in doing so, making the world a better place.