Our work speaks for itself.
We showcase implementation work spanning different domains and different starting points. We show it through code, operations, and numbers — not claims.
* For confidentiality, we withhold client names, product names and some specifications, focusing instead on the challenge, design, implementation and operation.
- In-house product
Grift — an AI estimation support tool that turns estimate rationale into a company asset
A product under validation that does not hand estimation over to AI, but instead increases the rationale that people can take responsibility for. Challenge: in contract development, estimation know-how depends on individuals and the rationale easily becomes a black box. Design & implementation: AI organizes past data and assumptions, and we designed and built a UI/UX that makes them easy for people to review and correct. Operation: currently in Team Beta validation, aiming to turn know-how held in individual heads into a company asset and to run estimation transparently.
- OSS / Public release
Engineer Cafe Navigator
An AI guidance system that answers visitors' questions in natural dialogue based on facility information. Challenge: responding to the wide range of questions about the facility was costly, and visitors had difficulty finding information. Design & implementation: we organized public information and FAQs and designed and built an AI chatbot platform that generates appropriate answers. Operation: released as open source, aiming for a structure in which the community can lead continuous improvement and operation.
- AI platform development
Development of a generative-AI SaaS platform that handles images and text together
A SaaS platform that combines multiple AI models to automate content generation for each purpose. Challenge: producing content that combines images and text required a large amount of human effort. Design & implementation: we designed an API integration platform that unifies several multimodal AI models and generates output matching the requirements consistently. Operation: we provided an interface that end users can operate intuitively and built an operating structure that allows continuous feature expansion.
* For confidentiality reasons, we withhold information that could identify the project.
- Core system renewal
Renewal and operational design of a large-scale survey collection and analysis platform
A system renewal for processing and analyzing a large volume of response data securely and in real time. Challenge: the performance of the existing system had declined, and both the lag before analysis and the operational load were increasing. Design & implementation: we designed and reimplemented load balancing that withstands heavy access, together with secure data storage rules. Operation: we put operation logging and permission management in place and moved to a data management structure that reduces concerns when running it day to day.
* Under NDA, the client name, product name, and some specifications are presented in abstracted form.
- Core system renewal
Cloud migration of a legacy core system, aligned with existing workflows
Support for migrating from an aging on-premises environment to a cloud environment designed with operation in mind. Challenge: the existing workflows had grown complex, and a straightforward move to the cloud was expected to cause confusion on site. Design & implementation: before migrating, we put the operational issues on site into words and organized them, removed unnecessary functions, and then designed and built the cloud architecture. Operation: we clarified the scope of responsibility to check during operation and established maintenance and operation flows the team can follow without hesitation.
* For confidentiality reasons, we withhold information that could identify the project.
- PoC / 0→1 development
Proof of concept for AI that automatically generates architectural drawings and floor plans
A validation combining specialist knowledge from the architecture industry with AI to support drawing work in the early design phase. Challenge: preparing floor-plan proposals in the early design phase took time, which held back the speed of client proposals. Design & implementation: we defined data-shaping rules so that AI could work with building standards and industry-specific design rules, and designed and built a prototype. Operation: we set out a flow in which people carry out the final check of the generated drawings, and assessed whether it could be incorporated into day-to-day work.
* Under NDA, the client name, product name, and some specifications are presented in abstracted form.
See something that resonates?
Let us think through your challenges together.