Salesforce + AI field guides
Make the architecture decision before the software decision
Practical guidance for the people accountable for Salesforce, revenue, service, and AI—built around operating workflows, permissions, failure modes, and measurable work.
Start here
The decisions that shape the build
Product documentation explains a feature. These guides explain how the feature fits into a system people must operate.
9 min read
Salesforce AI Production Readiness: Permissions, Evaluation, Approval, and Rollback
A practical readiness framework for moving a Salesforce AI workflow from an impressive demo to governed production work.
Open the readiness framework →
Agentforce vs MCP vs RAG: Choosing the Right Salesforce AI Architecture
Separate the user interface, connection, retrieval, action, and governance layers before choosing a Salesforce AI platform.
Salesforce AI Chatbot Architecture: Grounding, Case Creation, and Human Handoff
How to design a customer, patient, or internal service chat that retrieves the right context, creates governed work, and hands risk to a person.
Start with the operating problem
Choose the decision you are making
AI + Salesforce
Adopt AI safely
Choose the architecture, permissions, evaluation, and operating controls behind useful AI—not another disconnected pilot.
Browse this track →
Revenue systems
Fix revenue handoffs
Design capture, routing, lifecycle, attribution, follow-up, and reporting as one observable operating motion.
Browse this track →
Service + case operations
Modernize service
Connect intake, triage, queues, knowledge, chat, documents, escalation, and human ownership around the case lifecycle.
Browse this track →
Platform transformation
Change the Salesforce core
Make sound choices across implementations, integrations, automation layers, managed ownership, and org change.
Browse this track →
Resource library
Guidance organized around the system
Use the architecture guides to choose a path, the playbooks to plan delivery, and the product guides when the connector itself is the decision.
13 resources
AI + Salesforce
How We Turned a Car-Rental Company's Entire Backend into AI Tools: A 114-Tool MCP Server in Production
How we wrapped a Caribbean car-rental company's entire backend in a 114-tool MCP server — SDK architecture, tiered write gating, and a public ChatGPT app.
AI-Assisted Salesforce Org Archaeology: Reverse-Engineering 216 Flows Before Touching Anything
An AI-assisted Salesforce org audit in practice: 5,200 metadata files retrieved, 216 flows mapped, and hidden routing conflicts found before any change.
From RAG Chatbot to Multi-Agent Salesforce Copilot: The ApexGenius Architecture
The real architecture behind ApexGenius: an n8n prototype, 12+ specialized agents, a 92,000+ chunk metadata RAG, and the production lessons in between.
A Voice AI Answers the Phone, Salesforce Gets a Perfect Lead: Wiring AI Phone Agents into CRM Intake
How an AI phone agent becomes a clean Salesforce lead: authenticated REST intake, platform-event buffering, and subscriber Flows from a live healthcare build.
Using Claude to Do Real Salesforce Admin Work: Permission Sets, Reports, and Routing Rules — With Guardrails
How a consultant runs real Salesforce admin work — permission sets, reports, Omni-Channel routing — through Claude via MCP, with a human approving every write.
ChatGPT as a Salesforce Admin Copilot: Case Rules, Omni-Channel Routing, and Reports
How Salesforce admins can use ChatGPT for case rules, Omni-Channel routing, and reports — through a governed MCP gateway with read-first, tiered write access.
Salesforce AI Production Readiness: Permissions, Evaluation, Approval, and Rollback
A practical readiness framework for moving a Salesforce AI workflow from an impressive demo to governed production work.
Agentforce vs MCP vs RAG: Choosing the Right Salesforce AI Architecture
Separate the user interface, connection, retrieval, action, and governance layers before choosing a Salesforce AI platform.
Salesforce MCP Implementation Cost: Connector, AI, Governance, and Delivery
Separate the AI, connector, Salesforce, and implementation bills—and budget the complete operating loop instead of one license line.
How to Connect Claude to Salesforce (2026): Step by Step
Connection paths, setup decisions, and governance controls for giving Claude useful access to an authorized Salesforce org.
Salesforce MCP Options Compared: Hosted MCP vs Agentforce vs ApexGenius vs Custom
Compare the main Salesforce AI connection paths across scope, governance, implementation depth, and operating ownership.
Agentforce Pricing Explained—and When MCP Does the Job
A decision framework for separating Salesforce agent platform costs from connector, workflow, and implementation costs.
Salesforce MCP Security: Identity, Permissions, Sharing, and Sensitive Data
What to inspect across the connector, Salesforce authorization, model context, tool scope, approvals, and audit path before launch.
3 resources
Revenue systems
One Org, Eight Downstream Systems: Designing an Integration Waterfall That Never Loses a Record
How one Salesforce org delivers qualified cases to 8+ partner law-firm systems — per-firm DTOs, Named Credentials, idempotent retries, and resend tooling.
Salesforce Lead Routing Architecture: From Form Fill to Accepted Owner
Design a Salesforce lead-routing system that captures demand, resolves identity, qualifies consistently, assigns deterministically, enforces acceptance, and exposes every failure.
How to Use ChatGPT with Salesforce: 7 Practical Workflows
Working patterns for revenue operations, support, administrators, and leaders—with the connection and governance choices made explicit.
5 resources
Service + case operations
Salesforce for Behavioral Health: What We've Actually Shipped (EHR Sync, Census Automation, AI Intake)
EHR sync on Platform Events, a self-recalculating daily census, clinical PDF generation, and voice-AI intake — four behavioral-health Salesforce builds with the real architecture and numbers.
We Built Real-Time Patient Chat Inside Salesforce with Platform Events (Instead of Buying Another Vendor Tool)
How we built real-time patient chat in Salesforce with Platform Events, empApi and CometD LWCs — 100,000+ messages managed, no separate messaging vendor.
Syncing an EHR with Salesforce in Real Time: The Platform Events Architecture That Kept Governor Limits Out of the Way
Platform Events architecture behind a bidirectional AdvancedMD–Salesforce sync: a 38-field patient event, subscriber Flows, MRN writeback, no trigger chains.
Who Viewed This Patient Record? A HIPAA-Style Access Audit Trail with One Platform Event and One LWC
Log who opened which patient record in Salesforce with one Platform Event and an invisible LWC — 34,860 audit events logged — plus the honest limits.
Salesforce AI Chatbot Architecture: Grounding, Case Creation, and Human Handoff
How to design a customer, patient, or internal service chat that retrieves the right context, creates governed work, and hands risk to a person.
4 resources
Platform transformation
Moving Real Money on Salesforce: Building Encrypted FedWire and NACHA Pipelines Inside Governor Limits
How we built FedWire, NACHA ACH, and Oracle GL pipelines on Salesforce: AES-256-CBC encryption in Apex, REST acknowledgments, and governor-limit engineering.
Stop Letting Webhooks Hit Your Triggers: Platform Events as the Shock Absorber for High-Volume Integrations
The Platform Events webhook buffer pattern from three production Salesforce builds: CTI dispositions, a 38-field EHR event, and 100,000+ patient messages.
AI-Augmented Salesforce Managed Services: What Changes—and What Still Needs Senior Ownership
A clear operating model for combining AI acceleration with fractional Salesforce administrators, developers, architects, and systems engineers.
n8n + Salesforce: 10 Automation Patterns—and When to Use Flow Instead
A practical guide to cross-system orchestration, failure recovery, and choosing the right automation layer for the job.
Product path
Need the Salesforce connector, not a consulting project?
ApexGenius connects authorized Salesforce orgs to Claude, ChatGPT, and compatible MCP clients for record, metadata, test, and approved deployment workflows. Use GAT when the job extends into architecture, custom systems, rollout, or ongoing ownership.
Initial systems consultation
Bring the operating decision. We’ll map the architecture around it.
- You leave with
- A focused consultation with a senior systems consultant
- A current-state fit and architecture read
- A recommended path — implementation, ongoing ownership, or product