The Marketing Department at ◯△× Co. manages its website using WordPress.com.
For articles requested by other departments, drafts are shared via a workflow application. These go through a structured publishing process that includes management and proofreading before publication.
However, when Marketing Department members write articles themselves, they log directly into WordPress.com to create drafts. Since these drafts don’t go through the workflow, the way they are handled is inadequate.
2. Solution: Centralize all drafts within a workflow system
To tackle this challenge, the process owner implemented a system whereby all drafts are managed within the workflow.
Specifically, a dedicated “Draft Submission” flow was added to the article creation process, establishing a mechanism for marketing team members to register their drafts.
As a result, both drafts requested by other departments and those created internally by the marketing team can now be centrally managed in the workflow system.
Before
View details of the workflow diagramBackup Requested Drafts
Drafts requested from other departments are automatically backed up.
Upload Images to WP
Media files are uploaded to WordPress.com.
Upload Draft to WP
An article (post) is created as a draft on WordPress.com.
Minor Edits and Publish
Marketing department members make minor adjustments to the draft and publish it.
Publication Notification
The article’s publication is shared internally within the department.
After
View details of the workflow diagramDraft Submission (Staff Only)
Marketing department staff can submit their created drafts here.
Backup Requested Drafts
Drafts requested from other departments are automatically backed up.
Proofreading
If all necessary items for publication aren’t set, a marketing team member proofreads the draft.
Upload Images to WP
Media files are uploaded to WordPress.com.
Upload Draft to WP
An article (post) is created as a draft on WordPress.com.
Minor Edits and Publish
Marketing department members make minor adjustments to the draft and publish it.
Publication Notification
The article’s publication is shared internally within the department.
Before/After(you can move the slider)
3. Benefits
Enhanced Article Progress Visibility
You’ll get a clear, centralized view of who’s writing which article, along with its progress and history.
Prevention of Missing Required Fields
The workflow app’s built-in checks eliminate input errors and omissions for essential fields.
Monitoring of Creation Volume
You can track the number of registered articles in real-time, providing an instant grasp of the workload.
4. Other Business Applications
This solution can be applied to tasks where it’s crucial to clarify who did what and to what extent, and to maintain a history of these actions.
Managing the production of company newsletters and internal communications.
Creating workflows for social media posts and campaign drafts.
Tracking the progress of manual and document creation.
Reviewing and publishing processes for video and design production.
Managing the review and approval of customer inquiry responses.
AI classifies inquiry urgency (A–E), notifies supervisors of critical cases, and enables faster responses.
1. Issue: The level of urgency is unknown
a BPO company that handles email inquiries on behalf of client companies, providing fast and efficient responses to customers.
The current inquiry process handles both low and high urgency requests identically, lacking a prioritization mechanism based on urgency.
It is becoming increasingly important to efficiently assess the urgency of customer inquiries and prioritize responses accordingly. For instance, if a zoo is the client, even critical issues like a lost admission ticket might be handled with the same priority as routine requests, hindering a quick resolution. This highlights the necessity of effectively determining the urgency of inquiries to ensure timely and appropriate responses.
The Process Owner addressed the problem by implementing an AI automated Step (AI Agent Task) and an automatic email sending step (Throwing Message Intermediate Event).
In the AI Automated Step, incoming inquiry emails are analyzed by AI. The AI assesses the email’s subject and text to automatically assign an urgency level, ranging from A (highest) to E (lowest), on a five-level scale. This assigned urgency level is then automatically added as a prefix to the beginning of the process subject.
Supervisors will receive automatic email notifications for inquiries classified as Urgency Level A.
Prioritizing urgent inquiries leads to faster responses, greater trust, and higher customer satisfaction.
Instruction for Emergency Level Classification to the AI Agent
An urgency level will be added to the beginning of the subject line.
Before
View details of the workflow diagram
Start process upon email arrival
50-character summary by AI agent
Email content is summarized
Draft reply text by AI agent
1. Create reply message
Send reply email
Request review from SV (Supervisor)
Draft reply text by AI agent
Send email
After
View details of the workflow diagram
Start process upon email arrival
50-character summary by AI agent
Email content is summarized
Draft reply text by AI agent
Urgency assessment by AI agent
Branching
If urgency is level A, a notification email is sent to the Supervisor
1. Issue: Staff knowledge is not being accumulated.
SakuSaku Mail is a BPO company that responds quickly to email inquiries on behalf of client companies.
The current inquiry handling process restricts input to assigned representatives and supervisors, limiting knowledge sharing from similar cases and insights from other departments. This lack of broader staff involvement results in delayed decision-making for complex issues and reduced response efficiency.
2. Solution: Share inquiry content in the internal chat
The Process-owner decided to incorporate the ability to post to the internal chat tool “Collab Chat” into the current process.
Specifically, a step will be added to the workflow that will automatically post the content of the email inquiry, its urgency, a brief summary, and a draft response to the company-wide chat.
Knowledgeable staff members can offer advice through chat comments, making it easier for inquiry representatives to gain new perspectives and receive advice based on past experiences from other staff members.
Before :
View Details
Process starts upon email arrival
50-character summary by AI agent
Email content is summarized
Urgency assessment by AI agent
Draft reply by AI agent
Branching
If urgency is level A, a notification email is sent to the Supervisor
1.Create reply message
Send reply email
Request review from Supervisor (SV)
Draft reply by AI agent
Send email
After :
View Details
Process starts upon email arrival
50-character summary by AI agent
Email content is summarized
Urgency assessment by AI agent
Draft reply by AI agent
Post to CollabChat
Branching
If urgency is level A, a notification email is sent to the Supervisor
1.Create reply message
Send reply email
Request review from Supervisor (SV)
Draft reply by AI agent
Send email
Compare Before/After
Before/After(you can move the slider)
3. Effects
Consolidation of collective knowledge via chat
Sharing information throughout the company has made it easier to gather knowledge from diverse perspectives.
Promoting communication among staff
This led to more active exchange of opinions and advice among staff members, which in turn led to more active daily communication.
Improved speed and quality of response
Complex inquiries can now be handled quickly and appropriately, improving client satisfaction.
4. Operations that can be expanded laterally
Handling inquiries from other channels
This approach can be applied to handling inquiries through multiple channels, such as phone, chatbots, and social media, enabling fast responses based on collective knowledge.
Internal FAQ and knowledge management
By accumulating the knowledge gathered among staff members and utilizing it in creating FAQs and manuals, we can promote standardization and efficiency in responding to inquiries.
Complaint and trouble handling
Even in cases where complaints or disputes are complex and difficult to judge, this system can be used to consolidate diverse opinions and past cases and make quick and appropriate decisions.
1. Issue: Delays in response due to inquiry confirmation work
SakuSaku Mail is a BPO company that provides initial responses to email inquiries received by client companies. With approximately 30 staff members, they handle a huge number of emails every day.
Prioritizing urgent inquiries is crucial, but currently, determining email urgency requires opening and manually checking each one. This time-consuming process delays responses to critical cases because selecting which emails to address first is inefficient.
2. Solution: AI-generated summaries in the subject line
A generative AI tool was implemented by the Process-owner to automatically summarize customer inquiries and use these summaries as the case names.
Upon receiving an email, AI generates a concise summary (under 50 characters), which is then automatically used as the case name.
Data Items and Task Processing Screen
My Task Page of Staff
Before
View Workflow Diagram DetailsInquiry Reception
When an inquiry email is received from a client company, the process begins.
1. Create Response
Staff create a reply to the inquiry email.
Send Response
A reply email is sent to the client company.
After
View Workflow Diagram DetailsInquiry Reception
When an inquiry email is received from a client company, the process begins.
50-Character Summary
Using generative AI, the content of the inquiry email is summarized into 50 characters.
Subject Line Update
The summary of the inquiry email is set as the task title.
1. Create Response
Staff create a reply to the inquiry email.
Send Response
A reply email is sent to the client company.
Before/After(you can move the slider)
3. Effects
Reduces the time required to confirm inquiry details
AI summarization significantly reduces the effort and time required to review content.
Accelerates emergency response
The summary display makes it easier to determine and respond to high-priority issues.
Reduces the burden on staff
By reducing the workload of checking inquiry details, the work load will be lightened.
4. Other business applications
Summary of internal reports and minutes
Quickly understand the main points of long reports and meeting minutes by summarizing them.
Organizing chats and messages
Summarize your daily chats and messages and share important information concisely.
Summarize customer interaction history
You can summarize past interactions briefly and smoothly transfer the content the next time you respond.
Add a supervisor advisory flow so uncertain AI-drafted responses are professionally reviewed, reducing errors.
1. Issue: Generative AI carries a risk of producing inaccurate responses
A BPO service provides initial responses to inquiries on behalf of client companies. Staff members create replies based on AI-generated draft answers, but these sometimes contain errors that staff members fail to notice before sending them to clients. Consequently, clients complain about receiving incorrect answers approximately once per week. This indicates a need to improve the accuracy of AI-generated drafts and the review process before responses are sent.
2. Solution: Add an Advisory Flow
To prevent AI from providing incorrect answers, the Process-owner has implemented a new advice flow. Now, after drafting a response, staff members who are uncertain about its accuracy can request a supervisor’s review. This step ensures an additional layer of verification before the response is finalized.
Professional review reduces ambiguity and improves answer accuracy.
At FukaFuka Building Materials Co. Ltd., the sales department’s core daily operation is responding to quotation requests for building materials. The 10-person sales team is tasked with providing prompt responses to diverse clients, ranging from major construction companies to local contractors.
*This process improvement story is fictional. It is not related to any real individuals or organizations.
1. Issue: Reduced Business Speed Due to Uniform Approval Process
In the sales department, junior employees are responsible for quotation requests that include a product model number, while senior employees handle requests where no product model number is provided.
During the approval process, the Sales Manager can easily approve quotation requests with model numbers, as these require no technical judgment. Conversely, requests without model numbers demand proposals for optimal products, necessitating technical skill and careful consideration.
However, while responsibilities are divided by the presence of a model number during quotation pricing, this distinction disappears during approval. All requests are processed through the same approval flow. As a result, the Sales Manager has to review every quotation request in detail. This leads to delays even for requests that could be easily approved, ultimately reducing the overall response speed.
2. Solution: Splitting Approval Processes Based on Presence of a Model Number
To address this issue, the process owner split the approval processes for simple quotations and advanced quotations.
Specifically, quotation requests with product model numbers (simple quotations), which were previously handled by a single approval process, were separated from those without model numbers (advanced quotations). This change allows the Sales Manager to identify the quotation type before approval. As a result, they can instantly determine the nature of the quotation and proceed with approval after thoroughly verifying its technical validity.
Before
View details of the workflow diagram0. Quotation Request Received (In-person / Web)
Quotation requests are received either in-person or via the web.
1a. Enter Quotation Price (with Product Model Number)
A sales team member (junior employee) enters the quotation price.
1b. Enter Quotation Price (without Product Model Number)
A sales team member (senior employee) enters the quotation price.
2. Quotation Approval
The Sales Manager reviews the quotation price and either approves or sends it back for revision.
2X. Revision Handling
If the quotation is sent back in step 2, the employee who initially entered the quotation price in step 1 makes the necessary revisions.
3. Quotation Submission Completion Report
Once the quotation is approved and submitted, a completion report is filed.
After
View details of the workflow diagram0. Quotation Request Received (In-person / Web)
Quotation requests are received either in-person or via the web.
1a. Enter Quotation Price (with Product Model Number)
A sales team member (junior employee) enters the quotation price.
2a. Quotation Approval (with Product Model Number)
The Sales Manager reviews the quotation price for requests with product model numbers and either approves it or sends it back for revision.
2Xa. Revision Handling (with Product Model Number)
If the quotation is sent back in step 2a, the employee who initially entered the quotation price in step 1a makes the necessary revisions.
3a. Quotation Submission Completion Report (with Product Model Number)
Once the quotation is approved and submitted, a completion report is filed.
1b. Enter Quotation Price (without Product Model Number)
A senior sales team member enters the quotation price.
2b. Quotation Approval (without Product Model Number)
The Sales Manager reviews and either approves or sends back the quotation.
2Xb. Revision Handling (without Product Model Number)
If sent back, the person who entered the price in 1b makes revisions.
3b. Quotation Submission Completion Report (without Product Model Number)
After approval and submission, a completion report is filed.
Before/After(you can move the slider)
3. Benefits
Accelerated Quotation Approval
By dividing the approval process, simple quotation requests are now processed quickly, which has boosted the overall speed of the sales department.
Improved Decision-Making Quality
The Sales Manager no longer needs to spend time on simple quotation requests, allowing them to concentrate on those requiring advanced technical skills. This has led to more appropriate judgments and proposals, ultimately improving the accuracy of quotation requests.
4. Other Business Applications
Streamlining Customer Inquiry Handling
By categorizing inquiries into those requiring standard responses and those needing specialized judgment, and then assigning different personnel to each, you can achieve both rapid answers and appropriate resolutions.
Automating and Dividing Order Processing
Separating the process for routine orders from those requiring special consideration allows for increased order processing speed while simultaneously preventing errors.
Don’t Let Notifications Get Lost! Ensure Critical Information is Delivered with Notification Prefixes in Email Subject Lines.
At DonGan Manufacturing, the subject lines for both approved and rejected quote notifications were identical, so the employees faced a risk of misinterpreting email notifications for quotes, potentially moving forward with quotes that had been rejected. To prevent this, we implemented a system that automatically adds a “[Rejected]” prefix to the subject line of rejection notification emails. Now, recipients can immediately understand the quote status just by looking at the subject line, effectively preventing the misidentification of approved versus rejected quotes.
*This process improvement story is fictional and not related to any real individuals or organizations.
1. Issue: Overlooking Rejected Quotes
DonGan Manufacturing is a company with approximately 200 employees that specializes in the production and sale of industrial machinery. Within its quotation approval process, a dedicated staff member prepares a quotation based on customer requirements. This quotation is then reviewed by a manager before being sent to the customer. Furthermore, quotations exceeding ¥5,000,000 require final approval by the officer in charge, ensuring the submission of highly accurate quotations.
However, a recurring issue has been the identical subject lines for both approval and rejection notifications. This has led to instances where employees mistakenly assume a quotation has been approved and proceed with related tasks. Continuing with this process risks not only wasting time and resources, but also potentially compromises the approval process’s reliability.
2. Solution: Enhance Identification by Adding Notification Prefixes to Subject Lines
The process owner implemented a system that automatically adds a “[Rejected]” prefix to the subject line of rejection notification emails. This allows recipients to immediately determine the outcome simply by viewing the email’s subject line, effectively preventing the misidentification of approved versus rejected statuses.
Before
View details of the workflow diagram1. Quotation Creation
The responsible staff member manually inputs the quotation details.
x1. Quotation Revision
The quotation content is revised and resubmitted. If the quotation is to be discarded, “[Failed]” is automatically prefixed to its status, and the process ends.
2. Manager Review
The manager reviews the quotation content and returns it for correction if there are any deficiencies.
3. Officer Approval
The Officer in charge approves or rejects high-value quotations. The result is then communicated via email.
Approval Notification
An email is sent to notify that the quotation has been approved.
Rejection Notification
An email is sent to notify that the quotation has been rejected.
After
View details of the workflow diagram1. Quotation Creation
The responsible staff member manually inputs the quotation details.
x1. Quotation Revision
The quotation content is revised and resubmitted. If the quotation is to be discarded, “[Failed]” is automatically prefixed to its status, and the process ends.
2. Manager Review
The manager reviews the quotation content and returns it for correction if there are any deficiencies.
3. Officer Approval
The Officer in charge approves or rejects high-value quotations. The result is then communicated via email.
Approval Notification
An email is sent to notify that the quotation has been approved.
Rejection Notification
An email is sent to notify that the quotation has been rejected. This email’s subject line will automatically include “[Rejected]”.
3. Benefits
Zero Misinterpretation of Notifications
Now, rejection notifications for quotes are immediately identifiable, eliminating the risk of incorrect work proceeding.
Reduced Rework
Employees can now take appropriate action without misinterpretation, leading to a reduction in unnecessary work.
Decreased Administrative Burden
Administrators no longer need to follow up on misinterpretations of rejections, reducing the effort involved in the approval process.
*This process improvement story is fictional and not related to any real individuals or organizations.
4. Other Business Applications
Application and Approval Workflows
Prevent delays caused by misinterpretation by adding prefixes to approval and rejection notifications for expense reports and leave requests.
Customer Support
Avoid overlooked or missed responses by adding prefixes to email subject lines indicating the status of support tickets.
Project Progress Management
Prevent oversight of project progress by automatically adding prefixes to notification subject lines for task completion or deferrals.
Kyoto, Japan – Questetra, Inc. has released a new feature called “AI Agent Task” in its cloud-based workflow product, Questetra BPM Suite, enabling easy and practical use of generative AI within workflow applications.
With this feature, generative AI can be seamlessly incorporated into workflow apps. For example, by simply configuring an automated step as “Summarize inquiries concisely,” each inquiry will be automatically summarized.
Background: The Gap Between Expectations and Reality in Generative AI Utilization
Interest in generative AI continues to grow among companies, with many considering its adoption to reduce man-hours and operational costs.
However, actual implementation often stalls due to challenges such as:
A lack of personnel with knowledge and skills regarding generative AI.
Difficulty identifying concrete use cases
Overview of the New Feature: What is AI Agent ?
The AI Agent allows users to input natural words instructions, based on which the AI automatically executes processing. No specialized knowledge such as programming or API configuration is required.
Powered by Anthropic’s LLM model, Claude 3.5 Sonnet v2, the AI Agent delivers high comprehension and accuracy.
Example instructions include:
“Please summarize the inquiry concisely.”
“Translate this application document into French.”
“Classify the report contents into categories.”
The AI output can be used as-is or reviewed by a human in a subsequent step, ensuring both safety and accuracy.
＜Workflow Diagram with AI agent incorporated＞
In Questetra BPM Suite, it has been possible to incorporate generated AI such as OpenAI ChatGPT and Google Gemini into workflow applications, but API linkage settings and individual contracts with each generated AI provider were required. The new functionality is available at no additional cost for all editions of Basic, Advanced, and Professional.
Benefits of AI Agent Task × Flexible Workflow Design
AI Agent Task is designed to handle commonly recurring use cases in business such as translation, summarization, classification, and requirement extraction. Moreover, Questetra BPM Suite allows AI Agent tasks to be freely placed anywhere in the workflow, and this flexibility supports ease of deployment in the field.
Thus, field personnel can experience the tangible effects immediately by incorporating and trying it out without worrying how to use AI.
Comments from the Development Manager
Questetra, Inc. CTO Akihiro Hatanaka
The AI Agent is tailored to address the biggest challenge in the introduction of generative AI in the field; We don’t know how to utilize generative AI in our business. This feature allows AI to be integrated into the business process as if it is a member of the field.
Comments from Partner Company
Naoto Kume, CEO, Highsmart Corporation
We offer consulting services specializing in the medical industry. We are supporting DX promotion using BPMS, especially in administrative operations where operations are required to be carried out by a smaller number of people.
The AI Agent is very attractive in that it can incorporate the power of generated AI into business processes even in the medical industry, where so-called IT departments are not well organized. It can execute tasks simply by writing what you want to be done, thus accelerating the on-site improvement cycle significantly. We regard it as a big step forward in autonomous business improvement.
About Questetra BPM Suite
Questetra BPM Suite is a cloud-based business process management system (SaaS BPMS). You can develop and operate workflow systems (Workflow Apps) entirely through a web browser, without needing programming knowledge (coding skills). This allows business departments to lead continuous process improvements.
You can systematize various day-to-day operations such as approval flows, quotation submission processes, and inquiry handling. Advanced automation, such as “Requesting AI to draft a document” or “Saving files to Google Drive,” can also be implemented in No-Code.
Questetra, Inc. is an enterprise cloud computing company in Kyoto Japan, founded in 2008. We optimize the world’s Business Processes. Visit https://questetra.com/
Corporate Name: Questetra, Inc. (株式会社クエステトラ) CEO: IMAMURA Genichi Corporate Address: 206 Takamiya-cho Oike Bldg. 4th Fl., Nakagyo-ku, Kyoto 604-0835, Japan Capital Stock: 184,057,500 JPY Founded: April 1, 2008 URL: https://questetra.com/ Contact: pr@questetra.com
Questetra BPM Suite においては、従来より OpenAI ChatGPT や Google Gemini といった生成 AI をワークフローアプリケーションに組み込むことは実現可能でしたが、その際にはAPI連携の設定や、生成 AI 提供各社との個別契約が必須でした。今般提供を開始した本機能は、Basic / Advanced / Professional の全エディションにおいて、追加費用なしでご利用いただけます。
AIエージェント工程 × 自由設計ワークフローのメリット
「AIエージェント工程」は、翻訳・要約・分類・要件抽出など、業務で頻出する代表的なユースケースに対応しています。加えて、Questetra BPM Suite は、ワークフロー上の任意の位置にAI工程を自由に配置できる構造を持っており、この柔軟性こそが現場での展開のしやすさを支えています。
