AI checks sales emails for clarity and grammar before sending, improving reply rates.
1. Issue: Frequent Sending of Difficult-to-read Emails
A cloud service provider is struggling with its inside sales efforts. Their team contacts prospects via email to secure appointments, handling dozens of inquiries daily to obtain contact information for potential customers.
Sales representatives currently draft emails based on templates, customizing the content according to the prospect’s company size and industry, and then it is sent automatically by the system. However, this customization often leads to an increase in grammatical errors and typos. As a result, it is believed that the contents of the emails are frequently misunderstood, lowering the credibility of the information provided. This is reflected in a noticeable decrease in their reply rates.
2. Solution: AI Checks Email Content Before Sending
This company is changing its policy for sending emails to prospective customers. They’ll now only send emails if generative AI deems them suitable for dispatch. The process owner will add a step where generative AI checks the email content for readability and correct grammar. As a result, emails will only be sent if the AI determines there are no issues.
Generative AI improves email clarity and readability, increasing customer interest and reply rates.
Before
View details of the workflow diagram
The person in charge registers the prospect’s address and email content
The system sends the email content registered for the prospect’s address
The person in charge records reply outcomes from prospects, such as their level of interest in the service
The system waits for the “2. Record Reply” step for 10 days. If it remains unprocessed, “No Reply” is automatically recorded
After
View details of the workflow diagram
The responsible person registers the prospect’s email address and the email content
The system has generative AI check the email content for readability and errors
The system will send the email content registered to the prospect’s address if it receives a “no issues” result from the generative AI. If the generative AI returns a “typo” result, the process will be returned to the “1x. Contact Email Correction” step
The person in charge records reply outcomes from prospects, such as their level of interest in the service, during the “2. Record Reply” step
The system will wait 10 days for the “2. Record Reply” process to complete. If it’s still unprocessed, it will automatically record “No Reply.”
Error detection in the workflow notifies writers of spreadsheet errors, ensuring accurate article counts and reliable progress reporting.
1. Issue: Overlooking Automated Process Errors
A cloud SaaS vendor aims to publish 100 “how-to” articles on its product website to communicate product value to customers.
Our web production team is responsible for creating these articles, using a workflow app powered by BPMS. Once an article is complete, a notification is automatically sent to Collab Chat, our internal chat tool. This message includes the cumulative number of articles published, keeping everyone informed about our progress toward the 100-article goal.
The cumulative article count is calculated through an automated process: First, the workflow app adds article details (title and URL) to a Google Spreadsheet. Then, the Google Spreadsheet automatically calculates the number of rows (representing the number of published articles) using the COUNT function. Finally, the workflow app retrieves this count and embeds it into the Collab Chat message.
Our main issue lies in overlooking errors within this automated process. Although infrequent, errors can occur when article information (title and URL) is automatically added to the Google Spreadsheet. These rare but previously unaddressed errors led to inaccurate article counts being shared on Collab Chat. Such discrepancies can affect progress management and potentially undermine the web production department’s credibility.
2. Solution: Implementing Error Detection
The process owner configured the workflow to handle processing failures. Specifically, if an error occurs during the article information addition process, the token proceeds to an “error boundary event”. By assigning the next step of the error boundary event, writers are notified of the error.
In the next step (manual execution during error handling), the writer can then advance the token back to the article information addition process.
Adding an error detection feature ensures accurate published article counts by preventing inconsistencies during the article information process.
Before
View details of the workflow diagram
1-1. Create Article
2. Review
3. Publish
4x. Add Row to Spreadsheet
A new row is added to the specified Google Spreadsheet.
5x. Get Count
The number of published article rows calculated by the spreadsheet is retrieved.
6x. Publication Report (Share Count)
A message confirming the article’s publication, along with the retrieved count, is shared with the department via Collab Chat.
After
View details of the workflow diagram
1-1. Create Article
2. Review
3. Publish
4x. Add Row to Spreadsheet
A new row is added to the specified Google Spreadsheet.
4. Manual Execution on Error
If an error occurs in step 4x, processing this step will re-execute “4x. Add Row to Spreadsheet.”
5x. Get Count
The number of published article rows calculated by the spreadsheet is retrieved.
6x. Publication Report (Share Count)
A message confirming the article’s publication, along with the retrieved count, is shared with the department via Collab Chat.
Prompt changes stop AI misrouting vague emails, routing them to staff review and improving response speed.
1. Issue: Errors in AI-Driven Respondent Selection
A trading company handles a wide array of office essentials, including copy machines and printers. To streamline their office phone reception, the company adopted a call answering service. Calls to their main number are handled by an operator, and the outcomes are shared with the administration department via email. An AI-powered system has been integrated to further enhance efficiency by prioritizing calls and selecting appropriate respondents.
However, this system presents a persistent challenge. When operator-generated emails are addressed to a vague recipient address such as “To Support Staff” or “To HR Department,” the AI attempts to force-select a respondent from its list of respondents.
This often results in response requests being misdirected to entirely unrelated individuals. Crucially, when the AI incorrectly selects a respondent for a critical inquiry, it can significantly delay the request from reaching the appropriate person, leading to slow response times.
While a certain degree of AI error is unavoidable, there is a clear demand for improvements to prevent the occurrence of unnecessary operational burdens.
2. Solution: Modifying the AI Prompt to Output “No Respondent”
To mitigate erroneous AI-driven respondent selection, the process owner modified the AI’s prompt. Specifically, when an email from an operator contains a vague recipient designation like “To [Department] Staff”, the AI is now configured to output a hyphen (“-“), signifying that there is no respondent, rather than attempting to select a candidate from the list of respondents.
Furthermore, a mechanism has been implemented to automatically branch to the “1. Respondent Determination” step whenever the AI outputs a hyphen. This configuration ensures that emails with ambiguous recipient designations, particularly those containing critical inquiries, are automatically routed to a member of the administration department to manually select the respondent.
Preventing incorrect AI judgments in ambiguous cases shortens staff identification time and improves customer response speed.
Before
View details of the workflow diagramx1. Acquire Email Content
Upon receipt of an email with the “phone” label in Gmail, its content is read.
x2. Importance Assessment by AI
The AI analyzes the email content to determine if it is a sales inquiry. The data item “Judgment Result” is set to “not important” for sales inquiries and “important” otherwise.
x3. Acquire Respondent List
User information registered in the workflow platform (Respondent List) is retrieved.
x4. Respondent Determination by AI
The AI analyzes the email content and selects a respondent (email address) from the Respondent List obtained in “x3. Acquire Respondent List.”
x5. Set Respondent
Based on the respondent (email address) selected in the “x4. Respondent Determination by AI” automated process, the handler for the “2. Record Response Outcome” process is set.
g1. AI Judgment Result
The path is selected based on the value of the “Judgment Result” data item: – If “not important”, proceed to “1. Respondent Determination”. – If “important”, proceed to the email transmission event (“m1. Urgent Response Request”).
m1. Urgent Response Request
An email is sent to administration department members and the respondent determined in “x3. Respondent Determination by AI”, prompting them to handle the “2. Record Response Outcome” process.
1. Respondent Determination
An administration department member determines the “Respondent” based on the email content. The respondent determined in “x4. Respondent Determination by AI” is pre-filled.
2. Record Response Outcome
The respondent reviews the email content and records how the inquiry was handled.
After
View details of the workflow diagramx1. Acquire Email Content
Upon receipt of an email with the “phone” label in Gmail, its content is read.
x2. Importance Assessment by AI
The AI analyzes the email content to determine if it is a sales inquiry. The data item “Judgment Result” is set to “not important” for sales inquiries and “important” otherwise.
x3. Acquire Respondent List
User information registered in the workflow platform (Respondent List) is retrieved.
x4. Respondent Determination by AI
The AI analyzes the email content and selects a respondent (email address) from the Respondent List obtained in “x3. Acquire Respondent List.”
x5. Set Respondent
Based on the respondent (email address) selected in the “x4. Respondent Determination by AI” automated process, the handler for the “2. Record Response Outcome” process is set.
g1. AI Judgment Result
The path is selected based on the value of the “Judgment Result” data item: – If “not important”, proceed to “1. Respondent Determination”. – If “important”, proceed to the email transmission event (“m1. Urgent Response Request”).
m1. Urgent Response Request
An email is sent to administration department members and the respondent determined in “x3. Respondent Determination by AI”, prompting them to handle the “2. Record Response Outcome” process.
g2. Respondent Set?: The path is selected based on the value of the “Respondent” data item:
If “Respondent” is not set (i.e., “-“), proceed to “1. Respondent Determination”.
If “Respondent” is set, proceed to “2. Record Response Outcome”.
1. Respondent Determination
An administration department member determines the “Respondent” based on the email content. The respondent determined in “x4. Respondent Determination by AI” is pre-filled.
2. Record Response Outcome
The respondent reviews the email content and records how the inquiry was handled.
Adding an electronic signature step ensures certificate authenticity, prevents tampering, and improves trust with investors.
1. Issue: Absence of Electronic Signatures
An unlisted company that does not issue stock certificates certifies shareholder status by issuing a Certificate of Shareholder Register Information to shareholders. This task is overseen by the Administration Department, which creates a PDF certificate based on the information on the shareholder register and sends it to the shareholder.
Currently, certificates are not digitally signed, and some venture capitalists (VCs) have pointed out that the possibility of tampering cannot be ruled out. This situation not only increases the time and effort required to verify the information, but may also hinder the building of trust in capital alliances with VC firms.
2. Solution: Add an Electronic Signature Step
The process owner added an electronic signature step to the workflow, specifically, a step where an executive electronically signs the certificate after the management department has completed checking the certificate.
This allows the executive in charge to manually add a digital signature and ensures the certificate is issued in the format required by the VC.
Digital signatures enhance certificate authenticity, increasing trust with investors.
Before
View details of the workflow diagram
Shareholder requests via public web form
Certificate PDF generation
Certificate PDF automatically generated based on shareholder registry
1. Request acceptance judgment
Determine if there are any deficiencies in the request content
2. PDF double-check
Double-check performed by another member of the Administration Department
Email transmission
After
View details of the workflow diagram
Shareholder requests via public web form
Certificate PDF generation
Certificate PDF automatically generated based on shareholder registry
1. Request acceptance judgment
Determine if there are any deficiencies in the request content
2. PDF double-check
Double-check performed by another member of the Administration Department
Introducing a double-check system reduced certificate errors and improved accuracy and reliability in issuance.
1. Issue: Errors in Certificate Issuance
An unlisted company that does not issue share certificates instead issues certificates to its shareholders. After receiving a request to issue a certificate, the person in charge of the Administration Department extracts the necessary data from the shareholder registry, creates a PDF of the certificate, and sends it. This series of tasks was carried out by a single person.
Occasionally, errors were made at the time of issuance, resulting in incorrect information being included in the certificate. These problems were caused by inadequate data extraction and verification, as well as a checking system that depended on one person, resulting in a loss of trust.
2. Solution: Implement a Double-check System
The process owner decided to introduce a double-checking system to improve the certificate issuance process.
A step was added in the workflow app where another staff member reviews the certificate PDF after it is generated. This ensures that the PDF certificate is reviewed by two different team members.
A double-check system significantly reduced errors in certificate issuance.
Before
View details of the workflow diagram
A shareholder submits a request through a public web form.
Automated PDF Generation
The certificate PDF is automatically generated based on the shareholder registry
1. Request Acceptance Review
The request content is checked for any deficiencies to determine if it’s valid
Email Delivery
After
View details of the workflow diagram
A shareholder submits a request through a public web form.
Automated PDF Generation
The certificate PDF is automatically generated based on the shareholder registry
1. Request Acceptance Review
The request content is checked for any deficiencies to determine if it’s valid
2. PDF Double-Check
A different member of the administration department performs a double-check of the generated PDF
A gym boosted conversions by extending its no-fee offer via automated follow-up emails.
1. Issue: Insufficient Outreach to Potential Members
A membership-based gym chain with 30 locations across the country offers prospective members a trial program before they decide whether to sign up.While the studio gauges interest after the trial, few individuals sign up immediately. This has highlighted a need to boost their conversion rates.
To address this, the Kyoto branch independently launched a “No Enrollment Fee Campaign”. This initiative waived the enrollment fee for anyone who joined within seven days of their trial. The campaign successfully increased the branch’s membership conversion rate by 1.5 times compared to before its implementation.
However, a new challenge has emerged: this campaign doesn’t effectively reach potential members who don’t sign up within the initial seven-day window.
2. Solution: Automated Email for Extended Campaign Offers
First-First Fitness Studio’s Kyoto branch is supercharging its “No Enrollment Fee Campaign” to better engage potential members. They’ll launch a special email campaign targeting individuals who didn’t sign up immediately after their trial. This email will announce a seven-day extension of the “No Enrollment Fee” offer, giving them another chance to reconsider joining.
To make this happen, the process owner will revamp the workflow system. It’ll automatically dispatch the “Campaign Extension Email” seven days after a free trial if the prospective member hasn’t yet enrolled.
The plan boosts conversions by extending the “No Enrollment Fee” offer, giving trial members a second chance to join.
Before
View details of the workflow diagram
3.Membership Confirmation
We’ll confirm your enrollment after the trial on the same day
4.Enter Keyword (for Pending Membership)
The assigned trainer will input three keywords gathered from the prospective member
The AI will generate a thank-you message
5.Content Confirmation
The receptionist will review the generated text and make minor adjustments to the phrasing
A thank-you email containing feedback from the assigned trainer will be sent to the prospective member
6.Later Enrollment
We’ll process the membership application if the candidate wishes to join at a later date. This option automatically expires if 7 days pass from the trial date.
7.Feedback
The trainer who provided the free trial will check the membership outcome and enter their reflections
After
View details of the workflow diagram
3.Membership Confirmation
We’ll confirm your enrollment after the trial on the same day
4.Enter Keyword (for Pending Membership)
The assigned trainer will input three keywords gathered from the prospective member
The AI will generate a thank-you message
5.Content Confirmation
The receptionist will review the generated text and make minor adjustments to the phrasing
A thank-you email containing feedback from the assigned trainer will be sent to the prospective member
6.Later Enrollment
We’ll process the membership application if the candidate wishes to join at a later date. This option automatically expires if 7 days pass from the trial date.
Send campaign extension email
If 7 days have passed since the trial date, an email will be sent
7.Final Judgment
After the extension email is sent, we’ll process the membership if there’s an enrollment request
8.Feedback
The trainer who provided the free trial will check the membership outcome and enter their reflections
To ensure the AI functions correctly, we need to prepare accurate data.
1. Issue: The Heavy Burden of Updating Call Handler Lists
BizyBizy Corporation, a leading provider of office essentials like copiers and printers, has been actively enhancing its phone answering operations. They’ve adopted an answering service where operators handle calls to the main number, and the results are shared with the administrative department via email. Recently, BizyBizy further boosted efficiency by integrating an AI-powered system for automatic call assignee selection.
However, managing this AI system presents its own set of challenges. The AI relies on a pre-established “responder list” for its selections, but this list needs constant updates due to changes in staff because of new hires, departures, or transfers.
Maintaining this list has become a significant burden for the process owner. When updates are delayed, the AI can’t select the right person for calls, especially for unregistered employees. This means staff often have to manually handle calls to compensate for the AI’s selection errors, undermining the very efficiency the system was designed to achieve.
2. Solution: Automate List Updates Using Platform User Information
To tackle these challenges, the process owner integrated a system into the phone answering workflow that automatically generates the responder list based on the workflow platform’s user information.
This change ensures the responder list is always kept up-to-date.
Before
View details of the workflow diagram
x1. Acquire Email Content
When an email with the Phone Call label is received in Gmail, its content is read.
x2. Importance Determination by AI
The AI analyzes the email content to determine if it is a sales solicitation. The data item Judgment Result is set to “not important” if it’s a solicitation, and “important” otherwise.
x4. Responder Determination by AI
The AI analyzes the email content and selects a responder (email address) from a pre-prepared responder list.
x5. Set Responder
Based on the responder (email address) selected in the “x4. Responder Determination by AI” automated step, the processing handler for the “2. Record Response Outcome” step is set.
g1. AI Judgment Result
Depending on the value of the Judgment Result data item, the path is selected as follows:
If “not important,” proceed to the “1. Determine Responder” step.
If “important,” proceed to the email sending event (note*) “m1. Urgent Response Request.” *Accurately referred to as [Message Sending Intermediate Event (Email)].
m1. Urgent Response Request
An email is sent to administrative department members and the responder determined in “x3. Responder Determination by AI,” prompting them to process the “2. Record Response Outcome” step.
1. Determine Responder
Administrative department members determine the “responder” based on the email content. At this point, the responder determined in “x4. Responder Determination by AI” is pre-filled.
2. Record Response Outcome
The responder confirms the email content and records how they handled the inquiry.
After
View details of the workflow diagram
x1. Acquire Email Content
When an email with the Phone Call label is received in Gmail, its content is read.
x2. Importance Determination by AI
The AI analyzes the email content to determine if it is a sales solicitation. The data item Judgment Result is set to “not important” if it’s a solicitation, and “important” otherwise.
x3. Acquire Responder List
User information registered in the workflow platform is acquired (Responder List).
x4. Responder Determination by AI
The AI analyzes the email content and selects a responder (email address) from the responder list acquired in “x3. Acquire Responder List.”
x5. Set Responder
Based on the responder (email address) selected in the “x4. Responder Determination by AI” automated step, the processing handler for the “2. Record Response Outcome” step is set.
g1. AI Judgment Result
Depending on the value of the Judgment Result data item, the path is selected as follows:
If “not important,” proceed to the “1. Determine Responder” step.
If “important,” proceed to the email sending event (note*) “m1. Urgent Response Request.” *Accurately referred to as [Message Sending Intermediate Event (Email)].
m1. Urgent Response Request
An email is sent to administrative department members and the responder determined in “x4. Responder Determination by AI,” prompting them to process the “2. Record Response Outcome” step.
1. Determine Responder
Administrative department members determine the “responder” based on the email content. At this point, the responder determined in “x4. Responder Determination by AI” is pre-filled.
2. Record Response Outcome
The responder confirms the email content and records how they handled the inquiry.
You can move the slider
*This process improvement story is a work of fiction. Any resemblance to actual persons, organizations, or events is purely coincidental.
3. Benefits
Improved Accuracy
Because the responder list is always up-to-date, the AI can accurately select the appropriate responder. This prevents human error and manual mistakes, enabling precise task execution.
Expedited Customer Response
With the immediate selection of the correct responder, the speed of customer service improves. This leads to higher customer satisfaction and strengthens trust.
Cost Reduction
By reducing the labor and time spent on manual update tasks, personnel and operational costs are lowered. As a result, corporate profitability improves.
4. Other Business Applications
The mechanism of utilizing user information registered in a workflow platform can be applied to the following types of operations:
Information Security Training
After the training manager determines the training content and schedule, a mechanism can be built where users registered in the workflow platform (which can be limited to specific departments) are automatically assigned as handlers for the “Result Report” step.
This streamlines the management of training progress and prevents coordination errors among responsible parties.
Year-End Tax Adjustment
After the administrative department sets the year-end tax adjustment schedule, a mechanism can be built where users on the workflow platform are automatically assigned as handlers for the “Document Submission” step.
This mechanism facilitates the dissemination of submission deadlines and the tracking of progress, leading to increased operational efficiency.
Event Planning
After the content and schedule of internal events are determined, a mechanism can be built where users registered in the workflow platform (limited to specific departments if necessary) are assigned as handlers for the “Participation Confirmation” step.
This enables quick and reliable confirmation of participation, contributing to the smooth operation of events.
Surging orders overwhelmed rotating staff, so tasks are auto-reassigned after two hours to prevent delays.
1. Issue: Shipping Delays Due to Overwhelmed Staff
A traditional inn, known for its exceptional cuisine and prime location near major tourist attractions, is now facing a new challenge. The recent launch of their signature hot pot soup in vacuum-sealed frozen packs has been a huge success, leading to a surge in orders via phone, fax, and their website.
While their current system attempts to balance the workload by rotating shipping assignments among the team, the sheer volume of new orders means that assigned staff are frequently overwhelmed, causing significant backlogs.
2. Solution: Automated Task Handoffs Based on Staff Availability
The process owner implemented a system that automatically reassigns shipping tasks to the next available staff member, if the initially assigned person can’t complete the work within two hours. This ensures that tasks are handed off proactively before delays occur, as assignments are rotated sequentially within the shipping team.
Automatically reassigning handlers prevents backlogs and ensures on-time shipping.
Before
View details of the workflow diagram1. Order Information Entry (Manual)
The team manually inputs customer order information.
2. Primary Shipping Task (Main Handler)
The main shipping handler carries out the shipping tasks. Once complete, a shipping confirmation email is sent.
x1. Subject Line Configuration
The workflow system sets the subject line for communications, including the order ID and order date.
x2. Primary Handler Assignment
The workflow system designates the main shipping handler by rotating assignments among the shipping team.
After
View details of the workflow diagram1. Order Information Entry (Manual)
Staff members manually input order information.
2. Shipping Task (Primary Handler)
The primary shipping handler performs the shipping tasks. A shipping confirmation email is sent upon completion.
2′. Shipping Task (Secondary Handler)
The secondary shipping handler performs the shipping tasks. A shipping confirmation email is sent upon completion.
x1. Subject Line Setting
The workflow platform sets the subject line with the order ID and order date.
x2. Primary Handler Assignment
The workflow platform designates the primary shipping handler from the shipping team on a rotating basis.
x3. Exclude Primary Handler from Candidates
The workflow platform adds a rule to exclude the primary handler from further assignment considerations.
x4. Secondary Handler Assignment
The workflow platform designates a secondary handler, excluding the primary shipping handler.
AI automates thank-you messages, reducing staff workload and improving efficiency.
1. Issue: The Effort of Writing Thank You Notes for Visitors
A membership-based gym chain with 30 locations across Japan offers a successful free trial program that significantly contributes to new member acquisition. While member management is centralized, each branch handles free trial registration and management independently.
The Kyoto branch has implemented a unique workflow system that includes a web form for trial sign-ups and automated booking confirmation emails. After a free trial, reception staff ask visitors about their interest in joining. However, many attendees don’t commit on the spot and prefer to decide later. This makes sending a prompt thank you email crucial to keep their interest high.
It’s been observed that sending a personalized thank you email, including a message from the assigned trainer, on the same day as the trial has significantly improved visitor impression and led to an average 5% increase in membership conversion rates. The thank you email sending function is also integrated into the workflow system.
Despite its proven effectiveness, the current process of trainers individually crafting these thank you messages is time-consuming and labor-intensive, leading to increased overtime hours. Furthermore, this task presents a significant psychological burden for trainers who struggle with writing. Therefore, efficiency is urgently needed to maintain the positive impact on conversion rates without overworking staff.
2. Solution: Automating Thank You Message Generation with AI
The process owner will revamp the workflow to enable automatic thank you message generation using AI (ChatGPT).
Based on these inputs, the AI will automatically generate the thank you message. This change eliminates the need for trainers to compose messages themselves.
Automating thank-you messages reduces workload and stress for trainers, improving productivity and overall business efficiency.
Before
View details of the workflow diagram3.Membership Confirmation
We’ll confirm your membership after your trial session concludes today.
4.Post-Trial Thank You (If Membership is Pending)
The assigned trainer will enter feedback for the trial participant, limited to approximately 100 characters.
A thank-you email containing feedback from your assigned trainer will be sent to all visitors.
5.Delayed Membership Enrollment
Should a visitor express a desire to join at a later date, we’ll process their membership enrollment. (This option will automatically expire after a set period.)
Post-Trial Review
Trainers who provided the free trial will check the membership outcome (whether the visitor joined or not) and enter their reflections.
After
View details of the workflow diagram3.Membership Confirmation
We’ll confirm your membership after your trial session concludes today.
Keyword Entry (If Membership is Pending)
The assigned trainer will enter three keywords gathered from the visitor during the consultation.
AI will generate the thank-you message.Review Message
The reception staff will review the generated message and make minor wording adjustments.
A thank-you email containing feedback from your assigned trainer will be sent to all visitors.5.Delayed Membership Enrollment
Should a visitor express a desire to join at a later date, we’ll process their membership enrollment. (This option will automatically expire after a set period.)
Post-Trial Review
Trainers who provided the free trial will check the membership outcome (whether the visitor joined or not) and enter their reflections.
