Skip to content
Contact Us

Salesforce Agentforce Pricing: The Real Cost in 2026

Contact Us

Author

Siddharth Mittal

August 25, 2026

Introduction

Salesforce Agentforce has quickly moved from being an interesting AI capability to becoming a serious consideration for companies running Salesforce at scale. Businesses are looking at agents that can answer customer questions, work with CRM data, update records, summarize cases, assist employees, and take action inside existing business processes. Once an organization moves beyond a small proof of concept, one question becomes unavoidable: what will Agentforce actually cost? The answer is more complicated than a single license price. In 2026, Salesforce offers consumption-based and user-based options, and the final cost depends on how the agent is designed, how often it is used, what actions it performs, and what Salesforce products the company already owns. 

What Is Agentforce?

Agentforce is Salesforce’s platform for building and deploying AI agents. The important difference between an agent and a traditional chatbot is that an agent can be designed to do work, not simply return a response. Depending on the configuration, it can access Salesforce information, follow business instructions, invoke actions, work with flows, update records, and hand a request to a human when the situation requires it. 

Take a simple service example. A customer asks, “Where is my order?” A basic chatbot might provide a tracking link. An Agentforce agent can be designed to identify the customer, retrieve the relevant order, check its current status, and explain the expected delivery date. If the business process allows it, the agent could also create or update a case. The value therefore comes from completing a business task, not just producing a sentence. 

The Main Agentforce Pricing Models

Salesforce currently offers several ways to pay for Agentforce. The main consumption options are Flex Credits and Conversations. Salesforce also lists an Agentforce User License and broader Agentforce 1 editions. This flexibility is useful because not every company uses agents in the same way. A company running customer-facing conversations may have a different requirement from a company using agents internally to help employees complete CRM work. 

Salesforce currently lists Flex Credits at $500 for 100,000 credits. A standard Agentforce action uses 20 Flex Credits, which works out to about $0.10 per action. Salesforce also lists Conversations at $2 per conversation. The Agentforce User License is listed at $5 per user per month and requires Flex Credits. Agentforce 1 Sales and Agentforce 1 Service are listed from $550 per user per month when billed annually. These figures are useful for planning, but the customer’s actual contract and order form remain the final source for commercial pricing. 

How Flex Credits Work

Agentforce Flex Credits are based on the work an agent performs. Salesforce describes an action as a specific function executed by an AI agent, such as updating a record, answering a product enquiry, summarizing a case, or executing a custom prompt or flow. At the published standard rate, one action consumes 20 credits. With 100,000 credits priced at $500, a standard action has a reference cost of approximately ten cents. 

The important part is not the ten-cent figure by itself. The important question is how many actions are needed to complete a real business request. A customer may ask one question, but the agent might need several operations to answer it properly. It could identify the customer, find an order, check another record, apply a business rule, and update a case. The user sees one interaction, while the platform may perform several pieces of work.

A Conversation Is Not the Same as an Action

This distinction is important when preparing an Agentforce budget. A conversation describes the interaction with the user, while an action represents work performed by the agent. One conversation can involve several actions. If a customer asks why a refund has not arrived, the agent may need to locate the customer, find the transaction, check the payment status, retrieve the applicable policy, and update the service case. Counting only the conversation would not give a complete picture of the work being performed. 

For that reason, companies should avoid estimating cost simply by counting customers or conversations. A better approach is to take a real business process and map the actions the agent will perform. Once that workflow is understood, the organization can estimate monthly consumption and compare the result with the available pricing models.

Agentforce 1 Changes the Calculation

Agentforce 1 is a different way of looking at the investment. Salesforce positions Agentforce 1 Sales and Agentforce 1 Service as broader Salesforce editions with CRM, AI, data, and other capabilities included. Salesforce currently lists these editions at $550 per user per month when billed annually. 

At enterprise scale, that number becomes significant. One hundred users at $550 per month represents $55,000 per month, or $660,000 per year, before taxes and other commercial considerations. That does not automatically make Agentforce 1 a bad choice. If the organization needs the wider set of capabilities included in the edition, the package may provide better overall value than buying individual components separately. The correct comparison is therefore the total package against the company’s actual requirements. 

Existing Salesforce Licenses Matter

The current Salesforce environment should be reviewed before buying anything new. A company that already has Enterprise or Unlimited licenses may have options to add Agentforce capabilities without moving every user to a completely different edition. Salesforce provides several add-on and consumption options, depending on the product, edition, and use case. 

This is where procurement and technology teams should work together. Instead of asking only, “What is the price of Agentforce?”, the better question is, “What do we already own, what do we need, and which pricing model fills the gap?” The answer can be very different for a company that already has a mature Salesforce implementation compared with a company starting a new CRM and AI program. 

The Hidden Cost Behind Agentforce

The Agentforce charge is only one part of the overall investment. A production agent normally depends on the Salesforce environment around it. The implementation may involve Salesforce data, flows, Apex, integrations, Knowledge, security configuration, monitoring, external systems, and other platform capabilities. Some of those services may have separate licensing or consumption requirements. 

Data is especially important. An agent cannot provide reliable answers if the information it needs is incomplete, outdated, or spread across systems that it cannot access. If customer, product, order, or service information lives in multiple applications, integration work may be required before the agent can do its job properly. That engineering work is part of the real project cost even though it may not appear as an Agentforce license charge. For organizations that do not have the required internal expertise, Agentforce consulting services can also become part of the implementation budget.

Agent Design Directly Affects Cost

Agent design can have a direct impact on consumption. Two companies can automate the same business process and still use a different number of actions. One implementation may have a clear workflow with a small number of useful actions. Another may perform unnecessary lookups or repeated operations. When usage is metered, those design choices can eventually affect the bill. 

This is why Agentforce should be treated as an architecture decision as well as a licensing decision. The objective should not be to make an agent perform every possible task. The objective should be to design a controlled workflow in which the agent performs the work that actually creates value. A well-designed agent can also be easier to test, monitor, troubleshoot, and maintain. Organizations may also work with Agentforce consultants to optimize agent architecture, reduce unnecessary actions, and establish an efficient implementation strategy

Testing Is Part of the Real Cost

Testing an AI agent is more involved than checking whether the final answer looks correct. Teams need to verify data access, permissions, actions, record updates, business rules, escalation behavior, integration failures, incomplete inputs, unexpected questions, and security boundaries. They also need to check whether the agent behaves consistently enough for the business process it supports. 

Consumption should be considered during QA planning as well. Salesforce’s 2026 Flex Credit rate card lists different multipliers for some production and sandbox usage types. Standard and custom Agentforce actions are listed at 20 credits in production and 16 in sandbox. Standard and custom Voice Actions are listed at 30 credits in production and 24 in sandbox. A large automated test suite that repeatedly invokes agents can therefore create a different usage pattern from a small manual test.

Voice and Other Usage

Voice introduces another consumption category. Salesforce’s 2026 rate card lists standard and custom Voice Actions at 30 Flex Credits in production. Voice implementations may also involve speech-related services, so organizations planning a large voice deployment should model those requirements separately. The same principle applies to other metered Agentforce capabilities: understand the usage type before assuming that every interaction has the same price. 

A Practical Cost Example

Consider a company expecting 20,000 customer requests per month. If each request requires two standard Agentforce actions, the organization would generate about 40,000 actions each month. At the published reference rate of $0.10 per standard action, that would be approximately $4,000 per month, or $48,000 per year. This is a planning example rather than a guaranteed bill, because actual usage depends on activity and implementation. 

Now change only the agent design. Suppose the average rises from two actions to five actions per request. The same 20,000 requests would generate around 100,000 actions. At the same reference rate, that becomes approximately $10,000 per month. The number of customers did not change. The number of conversations did not change. The amount of work performed by the agent changed, and that is what moved the consumption. 

Is Agentforce Expensive?

There is no useful yes-or-no answer. Agentforce can be relatively inexpensive for a focused use case with limited activity and clear business value. It can also become a substantial enterprise investment when the agent handles large volumes, performs multiple actions per request, requires integrations, or is combined with broader Salesforce products. 

The common mistake when evaluating Salesforce Agentforce pricing is to judge the technology from one headline number. Looking at $0.10 per action and assuming Agentforce will always be cheap is misleading. Looking at a $550-per-user Agentforce 1 edition and assuming every Agentforce project will cost that much is equally misleading. The correct number comes from the business process, usage pattern, current Salesforce licenses, and commercial model. 

What Companies Should Calculate Before Buying

Before signing an Agentforce agreement, the company should first define the exact business problem it wants to solve. It should then estimate how many customers or employees will use the agent, how many interactions will happen each month, and how many actions the agent will normally perform for each interaction. A realistic pilot is often the best way to replace assumptions with actual usage data. 

The financial model should also include additional Salesforce consumption, integrations, testing, monitoring, implementation, and ongoing maintenance. Existing licenses should be reviewed at the same time. This gives leadership a much clearer view of total cost and expected value rather than treating Agentforce as an isolated AI purchase. 

Implementation and Maintenance

There is also a people cost. Someone has to define the agent’s purpose, configure its instructions and actions, connect data, establish security rules, test the workflows, monitor production behavior, and improve the solution as business requirements change. A small proof of concept may be manageable with an existing Salesforce team. A large enterprise rollout can require architects, developers, administrators, QA engineers, integration specialists, and business owners. 

Agentforce should also be treated as an ongoing capability rather than a one-time project. Salesforce configurations change, business processes change, integrations change, and users discover new ways to interact with the agent. Monitoring consumption, quality, failures, and business outcomes therefore becomes part of the operating model.

Start Small and Measure the Result

A practical way to control both risk and cost is to start with one narrow business problem. Customer order status, case summarization, internal knowledge questions, or a simple service workflow can be useful starting points because the expected outcome is easy to define. The company can measure how often the agent completes the task, how often a human needs to intervene, how many actions are consumed, and whether the process actually becomes faster or cheaper. 

Once those results are known, the organization can decide whether to expand. This approach is much safer than launching multiple agents across the company and trying to understand the cost later. It also gives leadership a stronger business case because the expansion decision is based on real consumption and measurable business results. 

Final Thoughts

Salesforce Agentforce pricing in 2026 is more flexible than a traditional per-user software model. Salesforce provides consumption-based options such as Flex Credits and Conversations, as well as user-based licensing and Agentforce 1 editions. The published reference points are $500 for 100,000 Flex Credits, 20 credits for a standard Agentforce action, $2 per conversation, and Agentforce 1 editions starting at $550 per user per month. 

But the headline price is not the real answer. The real cost is shaped by what the agent does, how often it is used, how the Salesforce environment is configured, what additional services are needed, and how much effort is required to test and maintain the solution. A simple knowledge agent and a complex service agent can have very different cost profiles even though both are called Agentforce. 

For companies evaluating Agentforce, the best approach is to start with the business problem rather than the license. Build a realistic pilot, measure consumption, understand the required architecture, review the existing Salesforce contract, and calculate implementation and operational effort. That gives decision-makers a much more realistic view of the investment. 

In the end, the Salesforce price page tells you the rate. Your business process determines the bill. The companies that understand that difference will be in a much better position to use Agentforce where it creates genuine value instead of simply adding another AI product to their Salesforce environment.

Pricing Note

Pricing figures are based on Salesforce’s published Agentforce pricing and 2026 Flex Credit rate card. Salesforce states that pricing and availability can change; customers should confirm final pricing and entitlements in their current Salesforce order form and commercial agreement. 

Author

Siddharth Mittal

Siddharth Mittal — QA Lead & AI Enabler Passionate QA professional specializing in scalable test automation and practical AI-powered solutions. Experienced in Salesforce testing, automation frameworks, and leveraging AI to build applications that enhance productivity and user experience.

Recent Posts

FAQ

FAQs

Salesforce Agentforce pricing depends on the pricing model and how the agent is used. Salesforce currently lists Flex Credits at $500 per 100,000 credits and Conversations at $2 per conversation. It also offers user-based options and Agentforce editions. The final cost can vary based on usage, features, existing Salesforce licenses, and implementation requirements.  

Salesforce currently lists Agentforce Flex Credits at $500 per 100,000 credits. A standard Agentforce action uses 20 Flex Credits, which works out to about $0.10 per action. However, the actual monthly cost depends on how many actions the agent performs and which usage types are involved.

Agentforce Flex Credits are used to pay for the work an AI agent performs. Each action, such as answering a question, updating a record, summarizing a case, or running a flow, uses a certain number of credits. Standard Agentforce actions currently use 20 Flex Credits, while Voice Actions use 30 Flex Credits. This means your total cost depends largely on how much work your agents perform.  

Flex Credits charge based on the actions an agent performs, while Conversations use a flat price per conversation. Salesforce currently lists Flex Credits at $500 per 100,000 credits and Conversations at $2 per conversation. Flex Credits can be used across different Agentforce use cases, while Conversations are mainly designed for customer-facing agents. Salesforce currently does not support using both pricing models in the same Salesforce org.  

No. Salesforce currently states that Flex Credits and Conversations cannot be combined in the same Salesforce org. Companies therefore need to choose the pricing model that best matches their Agentforce use case. Flex Credits provide more flexibility across different types of Agentforce usage, while Conversations provide a simpler per-conversation pricing model for supported customer-facing use cases.

You Have Questions,
We Have Answers

Talk to our experts today and explore how we can help you build a connected and efficient digital ecosystem.

Engineering enterprise intelligence through AI and Cloud excellence.

AI Solutions

Salesforce

Industries

Company

Copyright © 2026. All rights reserved.

Terms

Privacy

llms.txt