The 2026 Salesforce AI Architecture: A Practical Executive Blueprint
A concise blueprint for connecting Salesforce, Data 360, Agentforce, external models and cloud services without building an ungovernable estate.
Read articleCloud Group Journal
Field notes for the agentic enterprise.
Clear, executive-level guidance on Salesforce, Agentforce, data, applied AI and the architecture required to make them useful.
36 articles · August 2023—July 2026Filter by topic
The archive
Three years of practical perspective.
Every article is anchored to the platform and market context of its publication month, then written for decisions that still matter.
Showing 35 articles
From Chat to Delegation: Long-Horizon Agents Change Delivery Economics
Long-running agents can compress consulting and platform delivery work—if teams redesign review, evidence and accountability around them.
Read article
The Governed Multi-Agent Enterprise
A practical control model for multiple agents that share context, delegate work and act across Salesforce and the wider cloud estate.
Read article
Headless 360: Salesforce Without the Browser
Salesforce’s headless direction turns CRM capabilities into reusable services for agents, applications and ambient workflows.
Read article
Human-Agent Capacity Planning for Salesforce Teams
A planning model for balancing people, agents and platform work when digital capacity can scale faster than supervision and change.
Read article
The Enterprise Agent Platform Is an Operating System, Not a Model
As enterprise agent platforms mature, differentiation moves to context, actions, governance and the system for continuous improvement.
Read article
Spring ’26: From AI Features to an Agentic Revenue System
Spring ’26 connected agentic capabilities across the revenue lifecycle. The design challenge became preserving one accountable commercial process.
Read article
Opening Agentforce to Builders: When Expertise Becomes a Product
An open builder ecosystem creates leverage when specialized knowledge is packaged with evidence, controls and lifecycle ownership.
Read article
Data Cloud Became Data 360. Your Data Operating Model Should Mature Too
The name change signaled a broader role for governed enterprise context. The operating model must expand beyond CDP ownership.
Read article
Agentforce 360: Designing the Agentic Enterprise in Layers
Dreamforce 2025 moved the conversation from isolated agents to an enterprise platform. A layered architecture keeps that ambition governable.
Read article
How to Decide Which AI Pilots Deserve Production
A portfolio decision model for scaling the few Salesforce AI pilots that have evidence, operational fit and compounding value.
Read article
GPT-5 and the Case for a Multi-Model Salesforce Architecture
GPT-5 raised the capability ceiling. Enterprise architecture still benefits from routing work to the simplest model that can meet the requirement.
Read article
The AI Unit Economics Model Your Salesforce Roadmap Needs
A practical way to compare AI, automation and conventional delivery investments using outcome-level economics.
Read article
Agentforce 3: Observability Becomes a Board-Level Requirement
As agents scale, leaders need a view of behavior, cost, control and business impact—not another dashboard of model activity.
Read article
From Seats to Actions: Building an Agentforce Cost Model
A unit-economics framework for Agentforce as flexible consumption makes completed actions—not users—the planning unit.
Read article
MCP and the New Integration Boundary for Enterprise AI
What the Model Context Protocol changes—and what it does not—when Salesforce data and actions become tools for AI agents.
Read article
Agentforce 2dx: Designing Proactive Agents Without Losing Control
A control model for agents that initiate work from events and changing data rather than waiting for a user request.
Read article
Spring ’25: RAG, Hybrid Search, and the New Service Architecture
How retrieval quality became a first-class Salesforce Service Cloud design concern as agentic service moved toward production.
Read article
When Agentforce Moves Into Slack, Work Design Matters
Bringing agents into collaboration software can remove switching costs—or create a parallel operating layer with weaker controls.
Read article
Agentforce 2.0 and the Rise of Digital Labor Design
Agentforce 2.0 expanded the toolset. The strategic work became designing coherent portfolios of human and digital capability.
Read article
Testing AI Agents Is Not UAT With More Scenarios
Why agent testing needs simulation, behavioral evaluation and production feedback—not only scripted acceptance cases.
Read article
Going Live With Agentforce: The Production Readiness Checklist
A practical checklist for moving an Agentforce use case from a persuasive demo to a controlled, measurable production service.
Read article
Agentforce: The Operating Model Shift Behind the Product Launch
Dreamforce 2024 made agents tangible. The larger change was how companies would divide work, accountability and improvement.
Read article
Preparing a Salesforce Org for Autonomous Agents
The configuration, ownership and process work to complete before granting an AI agent meaningful autonomy inside Salesforce.
Read article
An Evaluation Scorecard for Customer-Facing Generative AI
A compact scorecard for deciding whether a Salesforce AI experience is ready to represent the company to customers.
Read article
Summer ’24: Build the AI Foundation While the Product Surface Moves
How Salesforce teams could keep momentum through fast product change by investing in durable data, action and evaluation foundations.
Read article
GPT-4o and the Multimodal CRM Moment
What native voice, vision and text meant for service design—and why the experience still had to begin with a governed workflow.
Read article
Zero Copy and the End of the Default ETL Reflex
A decision framework for using Salesforce zero-copy integrations without turning architectural fashion into a new form of complexity.
Read article
Prompt Builder Is Architecture, Not Copywriting
Why enterprise prompts belong in the governed Salesforce solution design, alongside data models, automation and access controls.
Read article
How to Run an Einstein Copilot Pilot That Produces Evidence
A production-shaped pilot model for Einstein Copilot that tests workflow value, grounding quality and user trust together.
Read article
A 90-Day Salesforce AI Roadmap for 2024
A focused plan to turn broad AI ambition into one production-shaped Salesforce outcome, with evidence leaders can trust.
Read article
The AI Governance Charter Every Salesforce Program Needed
A concise governance charter for moving enterprise AI from scattered experimentation to accountable delivery.
Read article
From Chatbot to Action Layer: Designing CRM Assistants That Can Do Work
The architecture that separates a polished conversational demo from a dependable Salesforce assistant capable of governed action.
Read article
Winter ’24: Release Readiness Became AI Readiness
A practical release-management model for the moment Salesforce changes began arriving with data, trust and AI implications attached.
Read article
Einstein 1 Changed the CRM Architecture Conversation
The important Dreamforce 2023 shift was not another AI feature. It was the convergence of data, metadata, automation and generative AI.
Read article
The Data Contract Before the AI Pilot
Why a small, explicit agreement about customer data was the highest-leverage prerequisite for Salesforce AI in 2023—and still is.
Read articleTurn perspective into movement