TL;DR: Salesforce lists Agentforce at $2 per conversation, but that number is true for one narrow pricing model. Real deployments involve Flex Credits or per-user licensing, plus Data Cloud, plus implementation, and the actual total varies significantly based on usage.
$2 per conversation is true, and also nearly useless as a planning number
Salesforce’s headline Agentforce price is $2 per conversation. That figure is accurate, and it is also only one of three pricing models Salesforce currently offers, applies only to customer-facing agents under a specific buying structure, and does not include several costs most production deployments actually need. If you are budgeting off that single number, you are likely underestimating the real cost by a wide margin.
The three pricing models, in plain terms
- Per-conversation: $2 flat per conversation, for customer-facing agents. Simple to understand, expensive at volume, and Salesforce’s own definition of what counts as one “conversation” has been a point of confusion for buyers.
- Flex Credits: A consumption-based model at roughly $0.10 per agent action, sold in blocks of 100,000 credits for $500. A typical service conversation involves 8-15 actions, which works out to roughly $0.80-$1.50 per conversation, cheaper than the flat rate at meaningful volume, but harder to predict before you have real usage data.
- Per-user licensing: Add-ons starting around $125-150 per user per month, or full Agentforce 1 Editions starting around $550+ per user per month. This model gives finance a predictable number to budget against, which is why Salesforce introduced it after enterprise pushback on unpredictable consumption pricing.
What the $2 headline number leaves out
A realistic Agentforce budget usually includes more than the conversation or action cost itself:
- Data Cloud is effectively required for any production-grade deployment, since it is what gives agents a unified view of customer data rather than whatever happens to be in the immediate record. List pricing runs around $108,000 per year for 10 million unified profiles, though many businesses already on Salesforce at scale may already have this.
- Knowledge base setup, if you do not already have a structured knowledge base for agents to draw from, typically runs $10,000-$30,000 to build.
- Implementation and agent configuration typically runs $20,000-$100,000 depending on how many use cases and how complex the agent logic is.
- Integration to non-Salesforce systems, if your agents need to reach an ERP, EHR, or core banking platform, adds a real integration layer cost on top of everything above.
The honest caveat
Agentforce can deliver real value, but the ROI math depends heavily on volume. At low conversation volume, the fixed costs, Data Cloud, knowledge base setup, implementation, can dominate the total and make the economics difficult to justify. At high volume, the per-conversation or per-action cost starts to look genuinely competitive against the cost of human agent time. Before committing to a pricing model, it is worth piloting on the per-conversation rate first, measuring real usage for a few months, and only then evaluating whether Flex Credits or per-user licensing would have been cheaper. Committing to a consumption model before you have real usage data is closer to guessing than budgeting.
What to do next
Do not budget off the $2 headline figure alone. Map out which of the additional costs above apply to your specific use case, particularly whether you already have Data Cloud and a structured knowledge base, since those two items alone can determine whether a deployment is a reasonable investment or a significant unplanned expense.
Celumai delivers fixed-price CRM implementations, data architecture, and Gen AI programmes for mid-market and enterprise businesses across the US, UK, Europe, and Asia-Pacific.
Work with Celumai
Need help implementing what you just read?
Book a free 30-minute assessment. We will review your current setup and tell you exactly what we would do. No obligation.
