Server-side Automation
Automate and eliminate human tasks to improve operational efficiency
A wide variety of automated steps (Service Tasks) can be seamlessly embedded between human decisions and actions. By simply placing and configuring icons within the workflow diagram, process owners can ensure that automated processing is executed each time a case (process instance) reaches the step.
For example, by placing AI integrations, automated calculations, or external system integrations before or after human tasks such as approvals and confirmations, organizations can eliminate dependency on individuals, accelerate processing speed, and reduce errors.
Embed AI Agent Steps
Multiple business data inputs can be referenced, and multiple data outputs can be updated within a single AI Agent step. Even complex instructions such as “extract and compare” can be configured within one AI Agent step.
Five Usage Patterns of AI Agent Steps
AI Agent steps in Questetra BPM Suite are utilized across a wide range of business processes, including customer inquiry handling, approval workflows, email newsletters and press releases, advertising creative production, internal help desk operations, and online demonstrations.
Based on actual usage, the roles of AI Agent steps can be broadly categorized into five types: Generation, Classification, Extraction, Transformation, and Comparison.
Generation Instructions
AI acts as a creator, generating artifacts from scratch.
Draft Generation
Based on a customer inquiry, three response drafts are generated. Human users only need to select and fine-tune the best one.
Comment Generation
Drafts for both approval and rejection comments are generated for approval requests, supporting quick managerial decisions.
Title Generation
Multiple email subject line ideas are generated from perspectives such as sincerity, expectation, and curiosity, helping improve open rates.
HTML Code generation
Rich HTML email content aligned with brand identity is generated for customer notifications.
Press Release Draft Generation
Media-ready announcement drafts are automatically generated from product specification data.
▶︎ Lowers the barrier for initiating downstream human tasks.
Classification Instructions
AI acts as a judge, classifying business data.
Proofreading
Detects typos and inappropriate expressions. If corrections are required, suggested revision instructions are also output.
Compliance Check
Evaluates advertising copy against regulations such as the Act against Unjustifiable Premiums and Misleading Representations and pharmaceutical laws, classifying results as Pass / Needs Revision / Not Acceptable.
Categorization and Tagging
Classifies inquiries into categories such as technical questions, quotation requests, or complaints, enabling automatic assignment to responsible departments.
Risk Assessment
Scores severity and urgency as High / Medium / Low, helping prevent delayed responses to potentially critical issues.
Sentiment Analysis
Determines customer sentiment (satisfied, dissatisfied, neutral) from survey responses and meeting notes, converting qualitative feedback into quantitative data.
▶︎ Reduces variability caused by human judgment and accelerates automated branching in processes.
Extraction Instructions
AI acts as a filter, extracting necessary information.
Field Extraction
Extracts company name, contact person, phone number, and preferred date from free-text inquiry emails and stores them in structured data fields.
Anonymization
Detects and masks personally identifiable information (PII) such as names, phone numbers, and addresses, enabling safe external data sharing.
Summary Extraction
Extracts key decisions and next actions from lengthy meeting minutes in bullet-point format, reducing reading time.
Prompt Generation
Generates prompts for image generation AI based on design guidelines, helping maintain brand consistency.
▶︎ Supports better understanding by both humans and systems in downstream steps.
Transformation Instructions
AI acts as a converter, transforming data into different formats.
Translation
Converts overseas inquiries into bilingual English-Japanese text, supporting multilingual operations.
Tone Adjustment
Transforms rough bullet-point notes into polished business documents, standardizing writing quality.
JSON Conversion
Converts natural language into machine-readable JSON parameters for use in subsequent API integrations.
Structuring
Transforms free-text data into structured formats such as Markdown tables or CSV for downstream processing.
▶︎ Facilitates understanding and usability in downstream human and system processes.
Comparison Instructions
AI acts as an inspector, comparing multiple datasets.
Difference Analysis
Compares original and revised documents and lists all changes, improving proofreading efficiency.
Quotation vs. Order Matching
Compares quotations and received purchase orders, identifying discrepancies in amounts, delivery dates, etc., preventing transaction issues.
Progress and Trend Analysis
Compares reports from different periods to identify delayed tasks or emerging trends.
Duplicate Detection
Compares new customer data with existing master data and calculates duplication likelihood scores, maintaining data cleanliness.
▶︎ Significantly reduces oversight from manual visual checks.
Embed Update Data Steps
This is the most fundamental and powerful type of automated processing. By placing and configuring Update Data steps, operations such as string concatenation and arithmetic calculations can be automated. Unlike AI-based reasoning, data is updated accurately based on predefined rules (formulas).
String Data Updates
The system combines multiple business data values and fixed strings to generate new text, helping prevent inconsistencies and improving readability in subsequent steps.
Automatic Process Title Generation
Generates unique titles such as “ABC Corporation: Quotation Request” from the customer name and case name.
Notification Message Creation
Composes email bodies by combining seasonal greetings, update dates, and update details.
Numeric Data Updates
The system performs real-time arithmetic operations using numerical business data.
Total Amount Calculation
Calculates total amounts from unit price and quantity.
Profit Margin Calculation
Calculates gross profit and profit margins from revenue and cost data.
Date / DateTime Data Updates
The system automatically calculates dates and times such as “X days later” or “X minutes earlier” based on specific reference points.
Automatic Deadline Setting
Sets a response deadline (e.g., 7 days later) based on the process start date.
Reminder Scheduling
Sets reminder timings, such as 24 hours before an event.
Selection Data Updates
The system automatically sets values for radio buttons or dropdown fields.
Automatic Status Update
Updates progress status (e.g., from “Unprocessed” to “In Progress”) upon completion of approval steps.
Automatic Category Assignment
Assigns categories such as “Important Client” or “Urgent” based on predefined conditions.
User / Organization Data Updates
The system automatically assigns responsible users or departments.
Assignee Assignment
Sets the responsible user based on email address data.
Department Assignment
Assigns responsible departments based on AI classification results.
Embed Built-in Automated Steps
These are ready-to-use standard components available with minimal configuration. No coding knowledge is required.
Data Processing and Calculation
String manipulation, numerical calculations, and date arithmetic.
Cloud Storage Integration
Automatically uploads files to Google Drive, OneDrive, Box, Dropbox, etc.
Cloud Data Processing
Updates Google Sheets data, creates draft posts in WordPress, etc.
PDF Generation
Automatically generates documents such as quotations and invoices by merging business data into predefined templates.
Integration with Generative AI Service
Custom integrations with OpenAI, Gemini, Claude, etc., using your organization’s licenses.
Embed Add-on Automated Steps
These automated steps can be installed by process owners (app administrators) or system administrators. With coding knowledge, custom automation components can be packaged and distributed internally.
Advanced Data Processing and Calculation
Regex-based replacement and extraction, TSV processing, master data updates, etc.
Advanced Cloud Data Processing
Overwrite TSV data in Google Sheets, convert Google Slides to PNG images, etc.
Embed Script Tasks
Simple data processing can be automated by configuring ECMAScript (JavaScript). Processing is executed by the GraalJS engine.
Text Data Formatting (Cleaning)
Removes blank lines from input text to maintain clean data.
// == Data Reference ==const strInput = engine.findDataByVarName( "q_Body" ); const arrInput = strInput.split("\n");// Remove blank lines and store in an arrayconst arrOutput = arrInput.filter(line => line.trim() !== "");// == Data Assignment ==engine.setDataByVarName( "q_Body2", arrOutput.join("\n") );
Extraction Using Regular Expressions
Extracts domains from email addresses and stores them for analysis.
const strEmail = engine.findDataByVarName("q_Email");// Capture domain part (after @) using regular expressionsconst match = strEmail.match(/@(.+)$/);if (match !== null) { engine.setDataByVarName("q_Domain", match[1]);}
BPM Platform Overview
Complete business process management lifecycle
