Everything you need before, during, and after your first Agentforce deployment.
The playbook collects the guides we use ourselves when scoping and delivering Agentforce work: a scoping checklist, two technical explainers, a worked case study, and the pricing and business case guides.
If you are new to Agentforce, start with how it works, which explains Atlas reasoning, Topics, Actions and the Einstein Trust Layer. If you are about to commission a first agent, read the scoping checklist. If you need to justify the spend, the pricing and ROI guides cover costs and how to build the case.
Salesforce changes pricing and product names often. Treat the cost figures as planning estimates and confirm against Salesforce's current pricing before you set a budget.
| Guide | Type | Read time | Read it when |
|---|---|---|---|
| How to scope your first Agentforce agent | Implementation guide | 8 min | You need to choose a use case and define data readiness, escalation and guardrails before building |
| Atlas Reasoning Engine explained | Technical deep dive | 12 min | You want to understand how agents classify topics and execute actions |
| A financial services Service Cloud build | Case study | 15 min | You want to see the architecture, MuleSoft integration and guardrails of a real deployment |
| Agentforce Pricing Guide 2026 | Pricing guide | 10 min | You are estimating per-conversation, Data Cloud, licence and implementation costs |
| The Agentforce ROI Guide | Business case | 9 min | You need a cost savings and revenue model to present internally |
| How Agentforce works | Technical guide | 14 min | You are a decision-maker who needs the full picture, including how it differs from a chatbot |
Salesforce admins, architects and business leaders who are evaluating, scoping or running Agentforce deployments.
Read "How Agentforce works" for the concepts, then "How to scope your first Agentforce agent" before you plan a build.
They are planning estimates written for 2026. Salesforce updates pricing and packaging often, so confirm against its current price list.
Yes. A discovery call can review your use case, data readiness and guardrail design.