Import or clone this Forge module into your factory after publishing the three Banking Agent modules it depends on. Your team configures AI models in each agent module, supplies documents and grounding data through the showcase UI, and uses the showcase to explore agent behavior. OutSystems does not manage your deployed copy.

• Banking Policy & Support Agent — exposes CallBankingPolicySupportAgent
• Banking Document Intelligence Agent — exposes CallBankingDocumentIntelligenceAgent
• Banking Risk & Fraud Assessment Agent — exposes CallBankingRiskFraudAssessmentAgent

1. Go to AI Models on ODC Portal.
2. Click on Add AI Model.
3. Add the available trial models on the Trial Models tab.

The landing page displays a card-based catalog under the heading Banking Agents Available in Forge. Each card contains:

• An icon representing the agent's domain.
• The agent name (e.g., Banking Policy & Support Agent).
• A short description of the agent's purpose.
• A category tag (Policy & Compliance, Document Analysis, or Risk & Fraud).

Clicking any card navigates to that agent's dedicated page.

Each agent page follows a consistent two-panel layout:

Header area:

• Back to All Agents link (returns to the landing page).
• Agent name and one-line description.
• View Agent on Forge button (opens the agent's Forge listing in a new tab).
• A note explaining that the interface is connected to the agent's public service action and that contractual/regulatory responsibility remains with the customer developer and operator.

Left panel — Parameter Reference:

• Input Parameters tab: lists each input (UserInput, SessionId, GroundingData, Documents, Profile, AgentDefinition) with its type and purpose.
• Output Parameters tab: describes the Response output, its format, and parsing guidance.

Right panel — Agent Interface:

• Agent status indicator (green dot when connected).
• Configure Agent button: opens a dialog to set Profile overrides (tone, verbosity, reasoning depth, risk tolerance, action autonomy, proactivity) and AgentDefinition extensions/overrides (Goal, Workflow, Knowledge, Tools, Guardrails, OutputDefinition).
• Document management bar: shows attached documents by name, with options to add or replace documents.
• Interface area (chat or otherwise).

Each agent page calls the corresponding agent's public service action on every user message:

1. User attaches banking documentation (fee schedules, account agreements, disclosures) via the document bar.
2. User types a question (e.g., "What are the overdraft fees for a checking account?").
3. The page calls CallBankingPolicySupportAgent with UserInput, SessionId, attached Documents, and any configured GroundingData, Profile, or AgentDefinition overrides.
4. The agent returns a Markdown document with sections (Answer, Key Terms, Rules, Rate & Fee Summary, Edge Cases, Conflicts, Not Specified, Assumptions, Regulatory References, Next Steps) and inline anchor citations.
5. The chat panel renders the Markdown response.

1. User attaches origination documents (applications, IDs, income proofs, bank statements) and provides grounding data (e.g., the Loan Application Object as JSON) via the document bar and GroundingData field.
2. User sends a message or instruction for the validation run.
3. The page calls CallBankingDocumentIntelligenceAgent with UserInput, SessionId, GroundingData, attached Documents, and any configured Profile or AgentDefinition overrides.
4. The agent returns a raw JSON object with Recommendation (valid/invalid), Title, Description, and DataSummary.
5. The Agent Interface panel displays the structured validation result.

1. User attaches case file materials (applications, IDs, income proofs, credit reports, statements, wire instructions) and provides grounding data (banking process type, known entities, thresholds) via the document bar and GroundingData field.
2. User sends a message or investigation focus.
3. The page calls CallBankingRiskFraudAssessmentAgent with UserInput, SessionId, GroundingData, attached Documents, and any configured Profile or AgentDefinition overrides.
4. The agent returns a raw JSON object with analysis_context, findings (severity-ordered), cross_document_consistency, credit_risk_indicators, recommended_detection_rules, and methodology_notes.
5. The Agent Interface panel displays the structured risk assessment.

The showcase demonstrates the recommended consumer integration pattern for all three agents:

1. Reference the Agent in your app's dependencies (ODC Studio > Add public elements).
2. Call the public service action (e.g., CallBankingPolicySupportAgent) from your screen action or server action.
3. Supply inputs: UserInput from user interaction, SessionId from your session management, Documents from your file upload, GroundingData from your business context, and optionally Profile and AgentDefinition for behavioral tuning.
4. Parse the Response: Markdown for Policy & Support; JSON for Document Intelligence and Risk & Fraud Assessment.
5. Render and act on the output in your UI or downstream logic.

This pattern is the same across all three agents — only the input content, output format, and business context differ. Use the showcase source as a reference implementation when building your own consumer modules.