Added Reconcilation Report Generation Fetaure #48

Adiking117 · 2025-10-19T13:41:28Z

📌 Title
Add Agentic Reconciliation Workflow with LLM‑Generated Reports and Email Delivery

📝 Description
This PR introduces a modular agentic workflow for automated account reconciliation in the banking portal. The workflow leverages LangGraph4j for orchestration and LangChain4j (Groq LLM endpoint) for intelligent report generation. It replaces ad‑hoc/manual reconciliation with a reusable, extensible pipeline.

🔄 Changes Introduced
Workflow Graph (GraphBuilder)

Defines a three‑node pipeline:

TransactionFetchNode → fetches transactions + user info

ReportGenerationNode → generates a natural‑language reconciliation report using Groq LLM

MailReportNode → emails the report to the account holder

LLM Integration

Configurable via application.properties (groq.api.key, groq.api.model, groq.api.endpoint)

Centralized LlmConfig bean for clean injection of OpenAiChatModel

Report Enhancements

Balance consistency check (opening + deposits – withdrawals – transfers = closing)

Narrative explanations in plain language (money‑diary style)

Alerts for large withdrawals and low balances

Weekly/monthly trend summaries

Professional, non‑technical formatting (no markdown, no IDs/codes)

Email Delivery

Uses EmailService to send reconciliation reports with a friendly, customer‑facing template

Controller

/api/workflow/run endpoint to trigger the workflow manually

⚙️ Configuration
Add the following to application.properties:

properties

LLM API

groq.api.key=your-groq-api-key
groq.api.model=openai/gpt-oss-120b
groq.api.endpoint=https://api.groq.com/openai/v1

✅ Benefits
Provides end‑to‑end reconciliation automation

Generates human‑friendly, diary‑style reports instead of raw technical logs

Modular design → easy to extend with fraud detection, compliance checks, or chatbot integration

Clean separation of concerns (fetch → generate → deliver)

🚀 Next Steps
Integrate with chatbot interface for on‑demand execution

Copilot

Pull Request Overview

This PR introduces an automated reconciliation workflow that generates AI-powered account reports and delivers them via email. The implementation uses LangGraph4j for workflow orchestration and integrates with Groq's LLM API (via LangChain4j) to generate natural-language, customer-friendly reconciliation summaries.

Key changes:

Agentic workflow pipeline with three sequential nodes: transaction fetching, LLM-based report generation, and email delivery
Integration with Groq LLM API for generating narrative-style reconciliation reports
New REST endpoint (/api/workflow/run) to trigger the workflow manually

Reviewed Changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 8 comments.

Show a summary per file

File	Description
application.properties.sample	Added configuration placeholders for Groq API credentials and endpoints
ReconState.java	Workflow state class managing account data, transactions, generated reports, and user information
TransactionFetchNode.java	First workflow node that retrieves transactions and user details for the logged-in account
ReportGenerationNode.java	Second workflow node that uses LLM to generate customer-friendly reconciliation reports
MailReportNode.java	Third workflow node that emails the generated report to the account holder
GraphBuilder.java	Orchestrator that builds and executes the three-node workflow graph
EmailServiceImpl.java	Added HTML email template method for reconciliation reports
EmailService.java	Interface method declaration for reconciliation report template
WorkFlowController.java	REST controller exposing the workflow execution endpoint
LLMConfig.java	Spring configuration for OpenAI chat model bean using Groq API settings
pom.xml	Added LangGraph4j and LangChain4j dependencies with version management

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

Copilot · 2025-10-22T13:57:34Z

src/main/resources/application.properties.sample

+groq.api.key = your-api-key
+groq.api.model = your-model
+groq.api.endpoint = your-model-endpoint


Property definitions should not contain spaces around the '=' sign. Remove spaces to follow standard Spring Boot property file conventions: groq.api.key=your-api-key

Suggested change

groq.api.key = your-api-key

groq.api.model = your-model

groq.api.endpoint = your-model-endpoint

groq.api.key=your-api-key

groq.api.model=your-model

groq.api.endpoint=your-model-endpoint

Copilot · 2025-10-22T13:57:35Z

src/main/java/com/webapp/bankingportal/workflow/state/ReconState.java

+    public static final Map<String, Channel<?>> SCHEMA = Map.of(
+            ACCOUNT_KEY, Channels.appender(ArrayList::new),
+            TXNS_KEY, Channels.appender(ArrayList::new),
+            REPORT_KEY, Channels.appender(ArrayList::new),
+            USER_KEY, Channels.appender(ArrayList::new)
+    );


Using appender channels for all keys creates unnecessary complexity since only the last value is retrieved. Consider using Channels.lastValue() or a single-value channel type for ACCOUNT_KEY, REPORT_KEY, and USER_KEY to simplify the retrieval logic and improve performance.

Copilot · 2025-10-22T13:57:35Z

src/main/java/com/webapp/bankingportal/workflow/nodes/TransactionFetchNode.java

+                ReconState.USER_KEY, Map.of(
+                        "name", userResponse.getName(),
+                        "email", userResponse.getEmail()
+                )


Type mismatch: USER_KEY is typed to store UserResponse objects (line 45-47 in ReconState), but this code stores a Map<String, String>. This causes a ClassCastException in MailReportNode line 26. Store the entire userResponse object instead: ReconState.USER_KEY, userResponse

Suggested change

ReconState.USER_KEY, Map.of(

"name", userResponse.getName(),

"email", userResponse.getEmail()

)

ReconState.USER_KEY, userResponse

Copilot · 2025-10-22T13:57:35Z

src/main/java/com/webapp/bankingportal/workflow/nodes/ReportGenerationNode.java

+        // Join the stringified transactions into one block
+        String txnSummary = String.join("\n", txns);
+
+        String prompt = """


Corrected spelling of 'Reconcilation' to 'Reconciliation' in the PR title (note: appears in title, not in this code file)

Copilot · 2025-10-22T13:57:36Z

src/main/java/com/webapp/bankingportal/workflow/graph/GraphBuilder.java

+        for (var item : compiled.stream(Map.of())) {
+            System.out.println("State after step: " + item);
+        }


Console output statements should be replaced with proper logging using SLF4J or similar framework. Add a logger instance: private static final Logger log = LoggerFactory.getLogger(GraphBuilder.class); and use log.info() or log.debug() instead.

Copilot · 2025-10-22T13:57:36Z

src/main/java/com/webapp/bankingportal/controller/WorkFlowController.java

+            graphBuilder.runGraph();
+            return ResponseEntity.ok("Workflow executed successfully");
+        } catch (Exception e) {
+            e.printStackTrace();


Using printStackTrace() is not recommended for production code. Replace with proper logging: inject a Logger and use log.error(\"Workflow execution failed\", e); to ensure exceptions are logged through the application's logging framework.

Copilot · 2025-10-22T13:57:36Z

src/main/java/com/webapp/bankingportal/service/EmailServiceImpl.java

 import java.io.File;
 import java.util.concurrent.CompletableFuture;

+import com.webapp.bankingportal.dto.UserResponse;


The imported UserResponse class is not used in this file. Remove this unused import.

Suggested change

import com.webapp.bankingportal.dto.UserResponse;

Copilot · 2025-10-22T13:57:37Z

src/main/java/com/webapp/bankingportal/service/EmailService.java


 import java.util.concurrent.CompletableFuture;

+import com.webapp.bankingportal.dto.UserResponse;


The imported UserResponse class is not used in this interface. Remove this unused import.

Suggested change

import com.webapp.bankingportal.dto.UserResponse;

Added Reconcilation Report Generation Fetaure

d21aef9

abhi9720 self-assigned this Oct 21, 2025

abhi9720 added the Feature Adds new functionality label Oct 21, 2025

abhi9720 requested review from AmrElsayyad and abhi9720 October 21, 2025 16:23

AmrElsayyad requested a review from Copilot October 22, 2025 13:55

Copilot AI reviewed Oct 22, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Added Reconcilation Report Generation Fetaure #48

Added Reconcilation Report Generation Fetaure #48

Uh oh!

Adiking117 commented Oct 19, 2025

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Oct 22, 2025

Uh oh!

Copilot AI Oct 22, 2025

Uh oh!

Copilot AI Oct 22, 2025

Uh oh!

Copilot AI Oct 22, 2025

Uh oh!

Copilot AI Oct 22, 2025

Uh oh!

Copilot AI Oct 22, 2025

Uh oh!

Copilot AI Oct 22, 2025

Uh oh!

Copilot AI Oct 22, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants


		import java.util.concurrent.CompletableFuture;

		import com.webapp.bankingportal.dto.UserResponse;

Added Reconcilation Report Generation Fetaure #48

Are you sure you want to change the base?

Added Reconcilation Report Generation Fetaure #48

Uh oh!

Conversation

Adiking117 commented Oct 19, 2025

LLM API

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Copilot AI Oct 22, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 22, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 22, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 22, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 22, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 22, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 22, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 22, 2025

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants