AI AGENTS
Development That Builds In AI Agents and Generative AI
AI produces results only when it keeps being used in daily work, not just when it is tried out. Here are cases that build in generative AI: an AI assistant that looks up data when you simply talk to it, AI scoring of personal statements, and automatic guidance right after an application.
Related Service: AI Implementation Support
CONSULTATIONS
Common Requests About Using AI in Daily Work
AI produces results only when it keeps being used in daily work, not just when it is tried out. Here are the requests we hear most often, and how we approach them.
Usage
We added an AI chat, but our work has not changed
Our Approach
A general-purpose AI chat is not connected to your business data or procedures, so the user has to bring in the information each time. As a result, it tends to end up as a tool that only some people use.
In our own business tool, AI agents with different roles work for each situation, such as estimates, progress management, meetings, and quality management, connected to the business data. We have also built an assistant that checks stock and a document-reading feature into our clients’ business systems.
A wrong answer from AI would be a problem
Our Approach
Generative AI can sometimes answer with plausible-sounding content that has no basis. If you use its answers in your work as they are, the errors reach your business partners and users.
We limit the information the AI is allowed to refer to, and we design it so that it does not make things up when it has no information. For scoring, the total score is not left to the AI but recalculated by a program, and bank passbook readings are imported only after the person in charge has checked them.
Safety and Cost
We do not want to send business data outside the company
Our Approach
When you hand customer information or internal documents to AI, you cannot decide whether to adopt it without knowing where they are sent and how they are handled.
In our own business tool, we let users choose, feature by feature, a setup that does not send business data to an outside AI provider. We use this mechanism and our operations know-how as they are in client projects too.
The more we use it, the more unpredictable the cost becomes
Our Approach
AI costs vary with usage. If you cannot see how much each feature costs, your anxiety grows the more you use it.
For the admissions matching service for universities, we record the cost of each use and route simple tasks to lower-cost models. In our own business tool as well, we manage operations by checking AI costs and the quality of results in numbers.
If AI goes down, will our work stop too?
Our Approach
An external AI service can become temporarily unavailable. If the workflow relies entirely on AI, work stops right along with it.
In the ordering and order management system and core business system for a construction company, every AI feature switches to regular processing when AI is unavailable, so work does not stop.
STRENGTHS
How We Can Help with AI Agents and Generative AI
We run AI in production in our own operations. The methods for building and running AI that we refined there are what we bring to your systems.
AI Know-How from Our Own Production Operations
In our project management tool, 45 AI features and the AI agent “Mr.AI” work every day across estimates, progress management, meetings, quality management, and multilingual support. AI agents handle the work from writing the specification for a change request through implementation, testing, and applying the change, and people check the results before they are incorporated.
AI That Works Inside Business Systems
We develop AI that is used within the flow of work rather than in a chat window: an assistant that looks up stock and orders when you talk to it, reading of PDF delivery schedules and bank passbooks, automatic sorting of site photos, notification summaries, and more.
CasesAI Assistant That Checks StockOrdering for Construction SitesCore Business System for a Construction Company
AI Features That Make Up the Service’s Value
After launch, we have added AI features that reach the users of a service directly, in stages: scoring of personal statements and essays, automatic generation of university information, writing of scout messages, and automatic generation of draft itineraries.
CasesUniversity Admissions MatchingAutomatic Itinerary Generation with AI
Managing Cost, Quality, and Safety in Operations
From the start, we build in the mechanisms that keep AI in use: recording the cost of each use, routing tasks to lower-cost models, limiting the information the AI may refer to, and switching over when AI is unavailable.
CasesUniversity Admissions MatchingInternal Business Platform with AI Built In
SCOPE
Features We Cover with AI Agents and Generative AI
We can add AI to an existing business system or build it into a new service.
AI Assistant
A feature that looks up business data and answers when you simply ask a question such as “How many orders did we receive today?”
Document Reading
A feature that reads PDF schedules, bank passbooks, handwritten answer sheets, and more, and imports them after confirmation.
Scoring and Feedback
A feature that scores personal statements and essays by criterion and returns strengths and points to improve.
Text Drafting
Drafting of page copy based on official information, scout messages, specifications, and more.
Answering Inquiries
A chat that answers questions from users and employees based on products, shipping, and internal records.
Automatic Sorting and Summarizing
Sorting of photos and documents, and judging the importance of notifications and summarizing them.
Work by AI Agents
A mechanism in which AI agents with different roles share the work from writing specifications through implementation, testing, and applying the changes.
Cost and Quality Management
Recording the cost of each use, using different models for different purposes, and switching over when AI is unavailable.
DIFFERENCE
The Difference Between General-Purpose AI Chat and AI Built into Business Systems
Both have their place. When you want to turn AI into business results, AI connected to your business data and workflow is the better fit.
| Aspect | Use a General-Purpose AI Chat | Build AI into Business Systems |
|---|---|---|
| Data Used | Users bring in the information each time | Connected to business system data |
| How It Is Used | Varies from user to user | Runs at set points within the workflow |
| Checking Answers | Users check for themselves | Limits the information referenced and builds in confirmation steps |
| Data Handling | Follows the service’s terms | Setups that do not send data outside the company can be chosen per feature |
| Cost | Often a flat fee based on the number of users | Records cost per feature and uses different models for different tasks |
FLOW
Our Process for Building AI into Daily Work
Rather than building big all at once, we build AI in one situation at a time, starting where it makes a difference.
Decide the Situations to Leave to AI
We sort out the workflow and decide in which situations AI can reduce effort and where people should check.
We decide first what to leave to AI and how to check it
Try It on a Small Scale
We try it on real data and check the quality of the answers, the cost, and the processing time.
Build It into the Business System
We build AI into the screens and workflow, and we also prepare a mechanism that switches to regular processing when AI is unavailable.
Expand While Watching Cost and Quality
We record the cost of each use and the quality of results, and expand features starting where the impact is greatest.
For the admissions matching service for universities, we added AI features in stages after launch
FAQ
Frequently Asked Questions About Using AI Agents and Generative AI
Here are the questions we are asked most often before a consultation. Please feel free to ask about anything not covered here.
Can we add AI to our existing business system later?
Yes. For the core business system of a furniture and interior wholesaler, we built in an AI assistant that looks up data and answers when you simply talk to it. For a construction company’s core business system, we have added bank passbook reading and sorting of site photos.
Case Study: Core Business System for a Furniture and Interior Wholesaler
Is it safe to give our business data to AI?
In the operations of our own business tool, we run a mechanism that lets users choose, feature by feature, a setup that does not send business data to an outside AI provider, and we design your project with the same approach. We develop under a framework based on ISMS (the international standard for information security).
Can AI give wrong answers?
Generative AI can make mistakes. We guard against them by combining mechanisms such as limiting the information the AI may refer to, designing it not to make things up when it has no information, redoing calculations with a program, and having a person check before anything is imported.
We are worried that AI usage fees will keep growing.
We record the cost of each use and route simple tasks to lower-cost models, so we run operations while keeping costs down. We also let you check in numbers how much each feature costs.
What happens to our work if an AI service goes down?
We build it to switch to regular processing when AI is unavailable. In the ordering and order management system and core business system for a construction company, this design keeps work from stopping.
What is an AI agent? How is it different from a chatbot?
A chatbot answers questions, whereas an AI agent is a mechanism that carries out work on its own within a defined role. In our own business tool, AI agents share the work of writing specifications for change requests, implementation, testing, and applying the changes, and people check the results before they are incorporated.
Where should we start?
We recommend starting with work that takes a lot of effort and whose correct answers are easy to check. Document reading and assistants that look up data and answer questions are areas where results are easy to see.
RELATED
Related Services and Fields
Case Studies by Field
All Case Studies31
Cloud Services9
Core & Business Systems4
Construction DX2
E-commerce Development3
Inventory & Order Management4
AI Adoption11
DX Promotion8
New Ventures12
Case Studies (11)
Cloud Services
AI Adoption
New Ventures
Nennai Nyushi Navi, a Matching Service for Comprehensive Selection and Recommendation Admissions: AI Feature Development and Ongoing Co-Creation
Nennai Nyushi Navi is a matching service that connects universities with high school students...
Core & Business Systems
E-commerce Development
Inventory & Order Management
AI Adoption
Moving Tens of Thousands of Items Correctly Across Multiple Sales Channels: A Core System and the AI Assistant "Buddio" for a Furniture and Interior Wholesaler
Cospa Creation Co., Ltd., a furniture and interior goods wholesaler that also sells directly...
AI Adoption
How a Company Changes When AI Is Built into the Core of Its Work: Running Our In-House Tool projectAI Company-Wide
We developed projectAI in-house: a project management tool designed on the international standard...
AI Adoption
New Ventures
Case Study: MEETSCUL, a "Hand-Raising" Matching Platform That Connects Regions and Companies with AI
When planning corporate retreats, offsite meetings or workations, companies traditionally had to...
Cloud Services
Construction DX
Inventory & Order Management
AI Adoption
Di-ORDER: Construction DX From Site Ordering to Core Operations, Built With a Team That Knows the Job Site
On a construction site, many parties exchange materials: tradespeople, general contractors,...
AI Adoption
DX Promotion
Case Study: A Membership Management System That Lets a Youth Soccer Club Handle Trial Applications and Enrollment on a Smartphone
A youth soccer club (a nonprofit organization) that has been active in Yokohama for more than 20...
E-commerce Development
AI Adoption
Case Study: An AI-Supported Shopify Store for a Furniture Seller That Sold Mainly Through Online Marketplaces
Cospa Creation Co., Ltd., which deals in furniture and interior goods, sold directly to consumers...
Core & Business Systems
Construction DX
Inventory & Order Management
AI Adoption
A Core System and Tradespeople App That Connect Every Part of a Construction Business: Atsuta Core System
At Atsuta Kenso Co., Ltd., an interior finishing contractor, ordering, crew assignment, invoicing,...
AI Adoption
Running Many Projects and AI Agents from One Screen and Your Phone: Our In-House Tool cockpit
cockpit is a tool we built ourselves to run AI agents working on many projects at the same time,...
AI Adoption
Running AI Without Data Leaving the Company: How We Choose Cloud AI or a Local LLM Feature by Feature
We run the AI features of our business tool, projectAI, by routing each feature either to "cloud...
AI Adoption
AI Takes Change Requests from Specification to Production: Running the AI Agent Mr.AI In-House
Our business tool, projectAI, has the AI agent "Mr.AI" built in. Mr.AI does more than answer...
FREE ESTIMATE
Free Estimates: Learn What to Build and the Cost Range on the First Call
Your requirements do not need to be settled. Tell us in a sentence what you want to build, and your project manager will sort it out with you.