From $2 Conversations to $0.10 Actions: Inside New Agentforce Pricing
When Agentforce first launched at Dreamforce ‘24, it promised a game-changing AI experience for enterprises. Designed to automate tasks across service, sales, and operations, Agentforce quickly gained attention for its advanced capabilities.
However, the excitement quickly faded due to one major challenge: pricing. At $2 per conversation, along with added complexities and hidden costs, Agentforce felt out of reach for most enterprises.
To solve these issues, Salesforce has updated Agentforce pricing model. The new approach replaces flat conversation fees with action-based pricing using Flex Credits, offering businesses a more affordable, transparent, and scalable way to adopt AI.
In this blog, we’ll dive deep into Agentforce’s revised pricing structure, why the change was necessary, and how it opens the door for more businesses to harness the power of AI effectively.
🤔Want to know what’s under the hood? Dive into our blog on how Agentforce works.
What to Expect in This Blog
Why Salesforce Revised Agentforce Pricing
The Original Pricing Model- How it worked and where it fell short
The New Flex Credit-Based Model- What changed and why it matters
A Direct Comparison- Old vs. New Pricing Model
Try Agentforce for Free
Other Major Salesforce Announcements
What Led Salesforce to Revise Agentforce Pricing?
At Dreamforce 24, Salesforce unveiled Agentforce. It instantly captured attention across the AI industry and was hailed as one of the most advanced AI technologies to date. And rightfully so, Salesforce had spent years investing in AI innovation, and Agentforce was positioned as the flagship solution, packed with cutting-edge capabilities.
But despite the initial hype, adoption progressed more slowly than expected, with many businesses hesitant to implement it.
Just three months later, in December, Salesforce released Agentforce 2.0, introducing major upgrades such as faster reasoning, Slack integration, and other key enhancements. While the 2.0 release addressed several gaps, the pricing model remained a significant barrier.
The original Salesforce Agentforce pricing model made it difficult for companies to forecast costs, with many finding it too expensive to justify, especially at scale. Over the past several months, both customers and Salesforce insiders raised concerns, prompting Salesforce to respond with a major pricing overhaul.
In fact, around 90% of CTOs shared that controlling AI expenses was a major hurdle, making it harder for them to unlock the full potential of Agentforce.
The Old Agentforce Pricing Model: What Made It Difficult to Scale
Initially, Salesforce introduced Agentforce with a conversation-based pricing model that quickly drew attention and criticism.
Each conversation was priced at $2 and defined as the period from the AI agent’s first response until the issue was resolved and closed by the customer, or after 24 hours of inactivity, whichever came first. If no interaction occurred for 24 hours, the conversation automatically ended, and any re-engagement started a new conversation, triggering another $2 charge.
Why This Was a Problem:
Unpredictable Costs: Businesses found it difficult to forecast expenses, especially for customer journeys that stretched over several days.
Increased Costs Due to Inactivity: Even simple follow-ups after 24 hours triggered a new conversation, driving up costs for unresolved or ongoing issues.
Flat Cost for All Use Cases: A simple query (like checking an order status) costs the same as a complex multi-step support case, which made the model feel unfair and inefficient.
Hindered Testing and Iteration: The high cost per conversation discouraged teams from experimenting with different AI workflows or refining Agentforce use cases during early adoption.
Conversation Lifecycle Example With Old Agentforce Pricing Model
June 1, 10:00 AM: A customer asks, “Can you update my shipping address?”
June 1, 10:01 AM: The AI agent responds and updates the address → $2 charged (conversation begins)
June 1, 10:30 AM: The customer asks, “Has the new address been saved?” → Still part of the same conversation
June 1, 11:45 AM: Customer confirms the update → Still within the same 24-hour window
June 2, 10:01 AM: Customer follows up: “Now, can you check my last order status?” Since more than 24 hours have passed → New conversation begins →Another $2 charged
While $2 per conversation may have sounded simple, the reality was more complicated. The $2 only covered the conversation itself, but Agentforce is more than just a chatbot. It’s designed to think, reason, and analyze data.
Whenever the AI agent needed to invoke an Einstein request for reasoning or retrieve specific data, additional charges were applied. In complex cases, where multiple reasoning calls were necessary, costs could quickly escalate. This made it difficult for businesses to accurately predict or manage the overall cost of using Agentforce.
The New Agentforce Pricing Model: Simpler and Predictable
To solve the challenges of the previous pricing and make Agentforce more accessible, Salesforce has transitioned from a conversation-based model to a flexible, action-based pricing structure.
Under the new Agentforce pricing model, each completed action costs $0.10 (20 Flex Credits) instead of $2 per conversation.
Let’s break down the key terms to understand how it works:
1. Action – An action refers to a task performed by Agentforce. This could include responding to a user inquiry or fetching data. Actions are counted once they are completed, meaning the AI has received a prompt, processed it, and delivered a response.
Examples of Common Actions and Their Credit Usage:
Update customer Record: 20 Flex Credits ($0.10)
Generate Case Summary: 20 Flex Credits ($0.10)
Search Knowledge Base: 20 Flex Credits ($0.10)
Execute Custom Flow: 20 Flex Credits ($0.10)
Schedule Appointment: 20 Flex Credits ($0.10)
Sandbox Environment: Agentforce actions performed in a sandbox consume 16 Flex Credits per task, offering a lower-cost option for testing and development.
2. Flex Credit- A flex credit is the core unit of this new Agentforce pricing model.
$500 = 100,000 flex credits
1 action = 20 flex credits = $0.10
Note on Token Usage:
Tokens are units of data processed by AI models and help measure the complexity and size of an action.
Each action includes processing of up to 10,000 tokens.
If an action exceeds this limit, it counts as multiple actions.
For instance, processing 20,001 tokens = 3 actions (charged as 60 Flex Credits).
Lengthy prompts or complex reasoning that cross token limits will be billed accordingly.
Salesforce Digital Wallet for Credit Management
To keep Agentforce usage predictable and efficient, Salesforce has introduced the Digital Wallet, a built-in tool that helps businesses track and manage their Flex Credit consumption in real time.
Key Features of the Salesforce Digital Wallet:
Real-Time Usage Tracking: Monitor how Flex Credits are being consumed,
Threshold Alerts: Set usage limits and receive proactive alerts when you're approaching your credit thresholds.
Usage Tagging: Gain detailed visibility with tags by environment, feature, or agent, helping you pinpoint where credits are being used most.
Trend Analysis: View historical usage patterns to forecast future credit needs and optimize your planning.
Old vs. New Salesforce Agentforce Pricing: A Side-by-Side Comparison
Understanding the shift from conversation-based to action-based pricing is key to estimating AI costs more accurately. Here’s how the two models compare:
| Feature | Old Model (Conversation-Based) | New Model (Action-Based) |
|---|---|---|
| Billing Basis | Per conversation | Per action |
| Base Price | $2 per conversation | $0.10 per action (20 Flex Credits) |
| Session Lifecycle | Session active until customer closes issue or 24 hours of inactivity | One complete task performed by the agent (e.g., update, summary) |
| Transparency | Difficult to predict the total cost | Clear and predictable credit usage |
| Usage Control | Limited | High control with Salesforce Digital Wallet |
| Cost for Simple Follow-ups | Can trigger a new $2 charge after 24 hrs | Counted as 1 action ($0.10), regardless of time gap |
Note: You cannot run both conversation-based and action-based (Flex Credit) pricing models in the same Salesforce org. You must choose one model per org.
Try Out AgentForce for Free with Salesforce Foundations
Salesforce is making it easier than ever to experience AgentForce at no cost.
As part of the Salesforce Foundations, businesses on Enterprise Edition or higher can now access 100,000 Flex Credits completely free. This is a golden opportunity to test AgentForce in real-world scenarios without worrying about upfront expenses or complex billing.
But that’s not all, Salesforce Foundations goes beyond Flex Credits. The plan also includes essential tools across Sales, Service, Marketing, Commerce, and Data Cloud.
Salesforce foundations | Source: Salesforce
Whether you're piloting AgentForce or exploring AI-driven automation across teams, this offer provides a safe, scalable way to get started.
Tip: Use these Flex Credits to evaluate specific use cases, fine-tune agent performance, and assess ROI before committing to a larger rollout
Other Major Salesforce Announcements
Alongside the updated Agentforce pricing model, Salesforce introduced two more key updates that give businesses greater flexibility and control when deploying AI at scale.
1. Flex Agreement: Convert Licenses ↔ Flex Credits
Salesforce now allows customers to convert unused user licenses into Flex Credits, and Flex Credits into User licenses. This new Flex Agreement gives businesses the agility to shift resources based on real-time needs.
Real-World Use Case: Seasonal Workflow Optimization
Peak Season (Q4 – Holiday Rush):
During the holiday quarter (October to December), businesses often experience a spike in customer service demand. Instead of spending more money, a company can:
Convert unused user licenses into Flex Credits
Use those credits to increase AI-powered support through Agentforce
Off-Peak Season (Q1 – January to March):
Once the holiday rush ends, support demand drops. The same company can then:
Convert the remaining Flex Credits back into user licenses
Use those licenses to onboard new team members or fill permanent roles
This dynamic license-credit exchange model is ideal for businesses with fluctuating staffing or workload cycles.
2. New Agentforce Licenses
To make it even easier for teams to work side-by-side with AI, Salesforce has launched dedicated Agentforce licenses. These are per-user, per-month licenses with no usage limits, removing the complexity of per-action billing for internal use.
This is ideal for employees who rely on AI agents daily for productivity, such as support agents, sales reps, and operations teams.
With these licenses, teams can focus on delivering outcomes without constantly monitoring credit usage.
Conclusion
The recent revisions to Salesforce Agentforce pricing make AI adoption more practical and budget-friendly. By shifting to action-based billing and introducing tools like the Digital Wallet and Flex Agreement, Salesforce has eliminated the unpredictability that previously held businesses back. Now, organizations can scale AI with greater confidence, cost control, and clarity.
Ready to scale with Salesforce AI? As a leading Agentforce implementation partner, Concret.io helps businesses integrate, customize, and maximize the value of Salesforce AI tools. Whether you're exploring Agentforce for the first time or expanding its use, we deliver tailored solutions that drive real results.
Suggested Read
Let’s Talk
Drop a note below to move forward with the conversation 👇🏻
