Integrations, Quality, and Delivery

Section 11: Use in Integrations

11.1 AI support areas

API design
payload mapping
JSON schema generation
authentication pattern explanation
named credentials planning
sequence diagrams
retry/error handling
logging strategy
event-driven architecture
webhook handling
middleware collaboration

11.2 Integration examples

#### Loan origination system

AI can help define:

application creation payload
underwriting update events
idempotent retry strategy
reconciliation rules

#### Payment gateway

AI can help define:

tokenized payment flow
safe error categories
PCI-sensitive fields to never expose

#### Identity provider

AI can help with:

SSO/SAML/OIDC explanation
user provisioning flow
role mapping model

#### ERP integration

AI can generate:

account/order/invoice mapping sheets
ownership matrix
sync frequency tradeoffs

#### Communication tools

AI can design:

SMS/email event flow
consent checks
communication logging strategy
campaign response integration patterns for marketing operations
audience segmentation data handoff guidance

11.3 MuleSoft and API-led guidance

Where MuleSoft or another middleware layer exists, AI can help define:

experience, process, and system API separation
ownership of transformation logic
replay and dead-letter handling
API versioning policy
support model between Salesforce, middleware, and source systems

11.4 Sequence diagram prompt

Create a text sequence diagram for Salesforce integrating with ERP for order fulfillment.
Include:
- order created in Salesforce
- middleware transformation
- ERP acknowledgment
- failure handling
- retry
- status update back into Salesforce

11.5 Integration checklist

What is the system of record?
Is the pattern sync or async?
How are retries handled?
Is idempotency required?
What gets logged?
Who owns schema changes?
What is the timeout strategy?
What is the support model?

Section 12: Use in Data Migration and Data Quality

12.1 High-value uses

source-to-target mapping drafts
cleansing rules
dedupe logic
validation planning
sequencing plans
mock datasets
UAT data strategy
data quality dashboards
reconciliation logic
cutover and rollback planning

12.2 Example prompt

Act as a Salesforce data migration architect.
Create a source-to-target mapping and migration checklist for:
- legacy customer table
- contact table
- loan table
- communication preference table

Target: Salesforce Account, Contact, Opportunity, custom Loan__c
Include:
- transformations
- default values
- dedupe rules
- validation rules impact
- load sequence
- reconciliation checks

12.3 Migration operating principles

migrate reference data before transactional data
define survivorship rules early
plan for inactive/archived records
separate cleansing from loading
define reconciliation reports before cutover

12.4 Data quality dashboard ideas

duplicate rate by source
mandatory field completeness
invalid email rate
orphan child records
status value normalization coverage

Section 13: Use in DevOps, Releases, and Environment Management

13.1 AI use cases

deployment plans
package manifest drafts
change risk summaries
release notes
regression scope identification
environment readiness checklists
user communication drafts
backout plans
branching strategy guidance
CI/CD explanation
Copado process documentation

13.2 Example prompt

Given this deployment scope, generate:
1. deployment checklist
2. pre-deploy validation
3. smoke test scope
4. rollback plan
5. support notification draft
6. release note summary for business users

13.3 Release manager checklist

metadata validated in lower environments
permission changes reviewed
data scripts sequenced
smoke tests agreed
backout criteria defined
support team briefed
user communication approved

13.4 Merge conflict help

AI can explain:

why metadata conflicts happened
which profile/permission set merges are risky
how to compare Flow versions safely
how to isolate high-risk deployment components

Section 14: Use in QA and Testing

14.1 Core uses

test scenario generation
test case generation
negative testing
edge-case identification
regression suite planning
UAT scripts
defect summary creation
traceability matrix drafts
risk-based testing plans
automation test ideas

14.2 Example scenarios

Process	QA Questions AI Can Help Generate
Lead conversion	What happens if duplicate account exists, required fields missing, owner inactive, or integration unavailable?
Contact creation	What if dedupe triggers, phone invalid, user lacks permission, or flow faults?
Case escalation	What if entitlement missing, queue full, SLA timestamp null, or omni-channel unavailable?
Approval process	What if delegated approver absent, approval recalled, or threshold changed mid-flight?
API failure handling	What if external timeout occurs, duplicate retry happens, or partial response returns?

14.3 Prompt example

Generate a risk-based test suite for Salesforce lead conversion.
Include:
- happy path
- negative path
- edge cases
- permission-based scenarios
- integration dependencies
- regression priorities
- suggested automation candidates

14.4 Human review required

QA leads must validate business risk ranking.
Test data assumptions must match actual org rules.
AI-generated test coverage is not a substitute for domain expertise.