IDFC+CROWAI.md

CrowAI + IDFC Bank: Authenticated Experiment Authoring Model

Problem

IDFC wants to run experiments on authenticated banking pages. However for the reasons below, it's not straight forward to setup these tests.

• production authenticated pages are reachable only by real customers
• no test or dummy accounts exist on production
• authors cannot log in to production for design-time inspection
• compliance and audit controls limit direct production access

Marketing teams need a safe way to create and validate variants (including custom JS) without production access.

Solution

Use UAT for authoring and validation, then promote only versioned configuration to production runtime.

• authors work on authenticated UAT pages with approved test credentials
• variants are created and tested against realistic non-prod flows
• validations run before promotion (selectors, safety checks, behavior checks)
• only signed/versioned variant config is promoted to production
• production snippet applies approved variants to real sessions in real time

Solution Sequence (UAT authoring, production runtime)

sequenceDiagram
  autonumber
  participant IDFC_Author as IDFC Experiment Owner (Growth/CRO)
  participant CrowAI_Editor as CrowAI Visual Editor (Authoring UI)
  participant IDFC_UAT as IDFC UAT Auth Pages (Test Credentials + Synthetic Data)
  participant CrowAI_Control as CrowAI Control Plane (Versioning/Validation/Promotion)
  participant IDFC_Prod as IDFC Prod App (CrowAI Snippet Installed)
  participant Bank_Customer as IDFC Logged-In Customer (Real Session)

  IDFC_Author->>CrowAI_Editor: Open experiment workspace and target page
  CrowAI_Editor->>IDFC_UAT: Load authenticated UAT page with approved test user
  IDFC_UAT-->>CrowAI_Editor: Return interactive page states for authoring
  IDFC_Author->>CrowAI_Editor: Create variant (UI edits + optional custom JS)
  CrowAI_Editor->>CrowAI_Control: Save selectors, mutations, audience rules, goals
  CrowAI_Control->>IDFC_UAT: Execute validation suite on authored variant
  IDFC_UAT-->>CrowAI_Control: Return pass/fail diagnostics and compatibility results
  IDFC_Author->>CrowAI_Control: Approve and promote validated experiment version
  CrowAI_Control-->>IDFC_Prod: Publish signed, versioned variant configuration
  Bank_Customer->>IDFC_Prod: Log in and navigate to target authenticated page
  IDFC_Prod->>CrowAI_Control: Request eligibility and variant payload
  CrowAI_Control-->>IDFC_Prod: Return targeting decision + mutation instructions
  IDFC_Prod-->>Bank_Customer: Apply variant in-session and continue journey

Loading

Assumptions

1. IDFC has a UAT/Test/Dev environment with authenticated pages that closely mirror production.
2. UAT test users and synthetic data are available and compliant for authoring use.
3. Authoring users (internal or agency) can reach UAT through allowed network paths.
4. UAT-to-production parity is strong enough for selectors and behavior portability.
5. UAT data covers major UI states seen in production journeys.
6. Selector contracts remain stable between authoring and runtime releases.
7. Custom JS behavior in UAT is a reliable predictor for production behavior.

Notes

• This model removes the need for production author login.
• It preserves production safety by promoting configuration, not direct manual edits.
• The highest ongoing risk is parity drift between UAT and production; this should be actively monitored.