Salesforce Agentforce Pricing Guide | SalesforceTutorial

Salesforce Agentforce Pricing: current list prices

Salesforce currently presents several ways to buy Agentforce. The public pricing page lists Conversations at $2 USD per conversation, Flex Credits at $500 USD per 100,000 credits, Agentforce User Licence at $5 USD per user per month when used with Flex Credits, and a no-cost Salesforce Foundations entry point for eligible Enterprise Edition and above customers. Salesforce also announced Agentforce add-ons starting at $125 USD per user per month and Agentforce 1 Editions starting at $550 USD per user per month. Check the official Salesforce Agentforce pricing page before quoting a customer or building a budget.

Agentforce $2 per conversation pricing

Agentforce $2 per conversation pricing charges a flat unit for the conversation model listed on Salesforce’s public pricing page. It is easiest to explain when the agent serves an external user through a support, sales, or service journey and the business already measures the value of each completed interaction. For example, if a support agent handles 8,000 billable conversations in a month, the simple list-price estimate is 8,000 × $2 = $16,000 before taxes, discounts, and contract terms.

Use this model only after defining what your business counts as a successful conversation. In enterprise orgs, the common mistake is to forecast total web chat sessions and forget containment rate, human escalation, failed authentication, bot loops, and repeated contacts from the same customer. Salesforce Agentforce pricing becomes easier to defend when each conversation maps to a case deflection, lead qualification, appointment booking, quote request, or other measurable business result.

Agentforce conversation pricing $2 per conversation: when it fits

Agentforce conversation pricing $2 per conversation can fit a narrow customer-facing use case where the outcome value is tied to the session. It is less flexible when the same agent needs to perform many small actions, call voice capabilities, use several prompts, or operate across employee and customer channels. Salesforce’s current pricing FAQ says Conversations and Flex Credits are not supported together in the same org, so architects should avoid designing a single org rollout that depends on switching models per agent.

A practical pilot should cap the audience, publish a clear fallback path to a human agent, and track the number of completed sessions daily. Do not launch a public agent and wait for the invoice to learn whether salesforce agentforce pricing matches your case deflection model.

Flex Credits in the Salesforce pricing model

The Salesforce pricing model for Flex Credits charges by credit consumption rather than by a flat conversation. Salesforce lists Flex Credits at $500 per 100,000 credits. Salesforce also states that each standard Agentforce action uses 20 Flex Credits and each Agentforce Voice action uses 30 Flex Credits on the public pricing page. At list price, that makes a 20-credit action equal to $0.10 and a 30-credit voice action equal to $0.15 before contract changes.

Flex Credits are usually the better model when the agent executes work: updating a record, retrieving knowledge, summarizing a case, running a prompt, invoking a flow, or calling a custom action. The forecast must count action executions. One chat that triggers eight actions can cost more than one chat that answers from a single knowledge retrieval. That is why salesforce agentforce pricing reviews should involve the Salesforce architect who understands the agent design, not only procurement.

Agentforce 1 Edition $550 per user month

The keyword agentforce 1 edition $550 per user month usually refers to Salesforce’s announcement that Agentforce 1 Editions start at $550 USD per user per month. Salesforce described these editions as generally available for Sales Cloud, Service Cloud, Field Service, and Industries Clouds, replacing earlier Einstein 1-style packaging with Agentforce-focused bundles. The announcement also says Agentforce 1 Editions include Agentforce add-on capabilities, 1 million Flex Credits per org per year, Data Cloud with 2.5 million Data Services Credits per org per year, and Slack Enterprise+.

Treat the Agentforce 1 Edition $550 per user month figure as a package comparison, not a standalone agent runtime rate. A CFO will want to know what existing Sales Cloud, Service Cloud, Data Cloud, Slack, Tableau, and AI add-ons can be retired or consolidated. An admin will want to know which users need the edition and which users only need access through another SKU. A solution architect should map the bundle to personas before choosing this path for salesforce agentforce pricing.

How does the Salesforce pricing model apply to agents?

The Salesforce pricing model has moved from simple seat planning to a mix of user licenses, usage units, and packaged entitlements. Agentforce adds a second question after “Who needs access?” The second question is “What will the agent do, and how often?”

For a normal Sales Cloud rollout, you can estimate user licenses by team. For Agentforce, you also estimate conversations, actions, prompts, voice usage, and the business event behind each interaction. This is why salesforce agentforce pricing should be reviewed with the business owner, Salesforce admin, enterprise architect, security lead, and finance partner.

Agentforce Builder, custom actions, and Agent Script influence cost because design choices change how many actions the agent executes. Salesforce Developer documentation explains that actions are the building blocks that let Agentforce agents perform tasks and interact with data, and Agent Script can control flow, variables, conditions, subagents, and action selection. See the Agentforce Developer Guide for the current development model.

How to estimate Agentforce cost before launch

Start the salesforce agentforce pricing estimate before you build the agent. The estimate does not need to be perfect, but it must show volume assumptions, unit price, expected business outcome, and a monitoring plan.

1. Classify the agent. Identify whether it is customer-facing, employee-facing, voice-enabled, or part of a packaged cloud edition.
2. Pick the pricing unit. Use conversations when the flat session unit matches the use case. Use Flex Credits when action execution is the main cost driver.
3. Estimate monthly volume. Use existing case, chat, lead, order, or call data instead of guessing.
4. Estimate actions per outcome. In Agentforce Builder, review how many actions the agent needs to answer, update, summarize, create, or escalate.
5. Apply list-price math. Use public list prices only for planning. Final pricing comes from your Salesforce agreement.
6. Track usage after launch. Salesforce Digital Wallet is the system Salesforce points customers to for usage-based tracking. Use it during pilot, not only after go-live.

Example monthly estimates

Apex cost estimator for admins

The following Apex class does not call Agentforce or replace Salesforce billing. It gives admins a bulk-safe calculator that can be exposed as an invocable action in Flow for planning workshops. Keep list prices in Custom Metadata in production so finance can update values without deploying Apex.

public with sharing class AgentforceCostEstimator {
    private static final Decimal CONVERSATION_USD = 2;
    private static final Decimal FLEX_CREDIT_BLOCK_PRICE_USD = 500;
    private static final Decimal FLEX_CREDITS_PER_BLOCK = 100000;
    private static final Decimal STANDARD_ACTION_CREDITS = 20;
    private static final Decimal VOICE_ACTION_CREDITS = 30;
    private static final Decimal AGENTFORCE_ONE_USER_USD = 550;

    public class Request {
        @InvocableVariable(label='Monthly Conversations')
        public Integer monthlyConversations;

        @InvocableVariable(label='Monthly Standard Actions')
        public Integer monthlyStandardActions;

        @InvocableVariable(label='Monthly Voice Actions')
        public Integer monthlyVoiceActions;

        @InvocableVariable(label='Agentforce 1 Users')
        public Integer agentforceOneUsers;
    }

    public class Result {
        @InvocableVariable(label='Conversation Cost Estimate USD')
        public Decimal conversationCostUsd;

        @InvocableVariable(label='Flex Credit Cost Estimate USD')
        public Decimal flexCreditCostUsd;

        @InvocableVariable(label='Agentforce 1 User Cost Estimate USD')
        public Decimal agentforceOneCostUsd;

        @InvocableVariable(label='Total Estimate USD')
        public Decimal totalEstimateUsd;
    }

    @InvocableMethod(
        label='Estimate Agentforce Monthly Cost'
        description='Estimates monthly Agentforce list-price cost for planning only. Use Salesforce Digital Wallet and your contract for billing.'
    )
    public static List estimate(List requests) {
        List results = new List();

        for (Request req : requests) {
            Integer conversations = req.monthlyConversations == null ? 0 : req.monthlyConversations;
            Integer standardActions = req.monthlyStandardActions == null ? 0 : req.monthlyStandardActions;
            Integer voiceActions = req.monthlyVoiceActions == null ? 0 : req.monthlyVoiceActions;
            Integer users = req.agentforceOneUsers == null ? 0 : req.agentforceOneUsers;

            Decimal conversationCost = conversations * CONVERSATION_USD;
            Decimal totalCredits =
                (standardActions * STANDARD_ACTION_CREDITS) +
                (voiceActions * VOICE_ACTION_CREDITS);
            Decimal flexCreditCost =
                (totalCredits / FLEX_CREDITS_PER_BLOCK) * FLEX_CREDIT_BLOCK_PRICE_USD;
            Decimal userCost = users * AGENTFORCE_ONE_USER_USD;

            Result output = new Result();
            output.conversationCostUsd = conversationCost.setScale(2);
            output.flexCreditCostUsd = flexCreditCost.setScale(2);
            output.agentforceOneCostUsd = userCost.setScale(2);
            output.totalEstimateUsd =
                (conversationCost + flexCreditCost + userCost).setScale(2);
            results.add(output);
        }

        return results;
    }
}

This code follows governor-limit basics: it performs no SOQL, no DML, no callouts, and processes every Flow interview input in a loop. If you adapt it to write estimates to a custom object, bulkify DML outside the loop and enforce object permissions before saving records.

How to choose the right Agentforce pricing option

Choose the model from the use case, not from the lowest list-price line. Salesforce Agentforce pricing looks simple only until an agent starts chaining actions, running prompt templates, calling flows, and handing off to users.

Use conversations when the business value is session-based

Use the conversation model when the agent handles a customer-facing journey where a session is the unit of value. Examples include case deflection, appointment booking, order status, policy questions, and lead qualification. The agentforce $2 per conversation pricing model is easier to explain to finance because the forecast is based on session volume, but it can become expensive when traffic grows without a matching business outcome.

Use Flex Credits when actions drive the outcome

Use Flex Credits when the agent performs work across systems or employee tasks. A service employee agent that summarizes cases, drafts replies, updates fields, and queries knowledge should be costed by action count. This is where the Salesforce pricing model rewards good design. Reducing unnecessary actions lowers spend without reducing access.

Use Agentforce 1 when the bundle replaces separate purchases

Agentforce 1 Editions make sense when the same users need the broader bundle and the included entitlements replace separate products or add-ons. The agentforce 1 edition $550 per user month search query should lead to a persona-based comparison: which users need the edition, which org entitlements are included, and which existing licenses become redundant.

Best practices for controlling Agentforce cost in production

Cost control starts in the design. A well-scoped agent has fewer billing surprises than a broad agent that can answer anything and call many actions. Use these controls before moving a pilot into production.

Design agents around measurable actions

Each agent should have a named business outcome: close a case, qualify a lead, book an appointment, update a record, or answer a knowledge-backed question. Avoid vague topics such as “help the customer” because they expand the action path and make salesforce agentforce pricing harder to forecast. Since Salesforce Developer documentation notes that Agentforce topics are now called subagents in newer documentation, keep naming consistent when you update build notes and runbooks.

Use Digital Wallet and contract guardrails

Salesforce states that Digital Wallet is available for tracking usage-based products, including Agentforce usage. Build a weekly review during pilot: compare forecasted conversations, actual conversations, forecasted actions, actual actions, and cost per resolved outcome. Ask your Salesforce account team which alerts, commitments, true-up terms, and overage rules apply to your contract.

Secure custom actions before scaling usage

Custom actions can create cost and security exposure at the same time. Salesforce Developer documentation says Apex methods annotated with @InvocableMethod can be made available as Agentforce custom actions. Before exposing an Apex action, use with sharing, validate inputs, enforce CRUD/FLS when reading or writing records, avoid SOQL in loops, and write tests for bulk inputs. See Salesforce’s custom Agentforce Apex action documentation before publishing a custom action to Agentforce Builder.

Separate pilot metrics from production forecasts

A pilot with internal testers rarely matches production traffic. Public users ask repeated questions, abandon sessions, fail identity checks, and return through different channels. For agentforce conversation pricing $2 per conversation, this can change the forecast quickly. For Flex Credits, production users may trigger different action paths than test users. Use pilot data to refine assumptions, not to approve an uncapped rollout.

Common errors with Agentforce pricing forecasts

• Counting users but not usage. Agentforce add-ons and Agentforce 1 Editions are user-based, but agents can still consume credits or other usage entitlements depending on the SKU.
• Ignoring action chains. A single user request can trigger several actions. Estimate action paths, not only the number of chats.
• Using list price as contract price. Salesforce public pricing pages are informational and subject to change. Contract terms decide the actual invoice.
• Mixing pricing models in one org design. Salesforce’s current pricing FAQ says Conversations and Flex Credits are not supported in the same org.
• Skipping security review. A poorly scoped action can update records too broadly and increase usage at the same time.
• Forgetting Data Cloud and prompt usage. Some Agentforce designs rely on data grounding, prompts, and related AI services. Include those usage types in the estimate.

Related SalesforceTutorial resources

For implementation context, read our guides to Salesforce Agentforce setup and architecture, Salesforce AI features for admins, Salesforce Data Cloud concepts, Salesforce Flow automation patterns, and Salesforce reports for usage monitoring.

Frequently Asked Questions