What you’ll find in this blog:
- An explanation of Agentforce’s consumption-based pricing model, a flexible alternative to traditional upfront license fees.
- A breakdown of why getting started with Agentforce may be more cost-effective and accessible than you think.
- Practical examples and use cases to help you estimate consumption and understand the different pricing structures.
Agentforce is evolving rapidly, with new releases and integrations landing almost every month. For consultancies and businesses eager to explore AI within the Salesforce ecosystem, it’s an exciting time.
So, what does it take to get started with Salesforce AI? The answer may surprise you.
Unlike many Salesforce products that require significant upfront license commitments, Agentforce uses a consumption-based pricing model. You pay only for what you use, billed in arrears each month. It’s a notable shift from the traditional Salesforce model that has been designed to make adoption more flexible, accessible, and cost-efficient.
Consumption Options
Before we delve into the differing consumption models, let’s just take a moment to understand the currency used within them, ‘FlexCredits’ & ‘Per Conversation Credits’.
FlexCredits
Salesforce has introduced FlexCredits as the variable currency behind Agentforce. Whenever you or your customers interact with an AI Agent, credits are consumed.
A single FlexCredit is priced at roughly £0.004.
Each agent interaction is measured in actions. On average, a typical request will take 3 actions to complete which includes things like retrieving a record, summarizing a pdf document or image, or finding the answer to a customer question on your website.
Every action consumes 20 FlexCredits, which works out at approximately £0.08 per action.
A little confused? Don’t worry. Basically, you can pretty much ignore the FlexCredit price, we really only care about actions as that’s where it becomes easier to estimate consumption and how it’s really measured.
We’ll explore in more detail below how actions are counted and why this model gives you a predictable, transparent way to manage costs.
FlexCredits are perfect for variable interaction with the AI agent that results in 7 or fewer actions when using list price.
Per Conversation Credits
Alongside FlexCredits, Salesforce also offers a per-conversation pricing model. You pay a flat rate of £1.60 per conversation.
This suits those who expect an internal task or customer interacting with your AI Agent to exceed 7 actions regularly. This gives you cost predictability without having to worry about actions.
Now that we have looked at the FlexCredit and Per Conversation pricing structures, let’s explore the options for applying them in your contract.
Consumption Models
There are three models available.
| Consumption Model | Pro’s | Con’s |
| Pay as you go | – Ideal for testing the waters with minimal risk – Full control over spend – No upfront commitment – Full access to all functionality – Paid in arrears based only on what you use | – Fixed pricing at £0.08 per action or £1.60 per conversation, no volume discounts – Costs can add up quickly at scale |
| Pre-commit | – Ability to negotiate lower per-action or per-conversation rates – No upfront payment (billed monthly against an agreed baseline) – Predictable budgeting (Agents can be turned off when limits are met) | – Tied into a contract/bill for the agreed term – Risk of overage charges if usage exceeds the baseline |
| Pre-purchase | – Maximum cost savings due to upfront commitment | – Full payment required upfront for the contract term – Still subject to overage if usage goes beyond what was purchased |
While the eye might immediately jump to the pay as you go model as the obvious choice, knowing your use case can give you good reason to consider the other models too:
- High-value transactions: If you operate in a sector like insurance or mortgages, where applications are multi-step, you often know your monthly volumes and can easily define the number of actions needed. A pre-commit or purchase model can significantly reduce costs in this type of use case.
- High-volume support: If you’re running a service desk or support queue, using an AI Agent to triage cases may require only a handful of actions per interaction, but at high volume. In that scenario, securing a lower per-action rate with FlexCredits could deliver meaningful savings.
Either way, if you’re just starting the journey, Pay-as-you-go remains the safest entry point. It lets you experiment without commitment, and once your consumption patterns are clearer and value is defined, you can confidently shift to Pre-commit or Pre-purchase to lock in better rates.
How do I measure my consumption and stop it getting out of control?
When you hear pay-as-you-go, it’s easy to think of a mobile phone plan where costs can spiral if you’re not careful. Thankfully, Salesforce have thought of this and provide the Digital Wallet.
When you activate Agentforce you get a digital wallet that allows you to monitor consumption in both your live and testing environments. Usage is refreshed regularly, so you always know where you stand. You can view and filter consumption by week, month, or year, giving you a clear picture of your upcoming bill and visibility into any potential overages across all models.
What’s more, agents can be individually activated or deactivated at any time, giving you additional control if you need to pause usage or manage spend against a budget.
How do I estimate my consumption and what will I pay? (Examples: Use Case & ROI)
To help you understand how you might estimate consumption we will use two illustrative use cases.
For reference, we’ve defined that an action costs approximately £0.08 and a conversation costs £1.60. Let’s now dive into the two use cases to show how actions are consumed and help you better understand potential consumption and ROI.
Use Case 1: Employee Onboarding Help (Internal) – Illustrative
Our use case involves an AI Agent answering employee questions about company policy, holiday allowance, company HR procedures. The AI Agent has access to the policy handbook, leave entitlements and HR records stored in Salesforce.
Employee: How many days annual leave do I have this year?
AI Agent: Your records show 25 days of leave entitlement, but you have 12 remaining. Would you like to see the leave days you have taken this year?
- The agent queries the HR record in context of the employee and responds accurately.
- Cost: 1 action = £0.08 + 1 satisfied employee.
The employee then asks:
Employee: Can I carry over unused holidays into next year?
AI Agent: The company policy allows you to take 5 unused days into the next year, (Section 5.2). Would you like me guide you through the process or create the carry-over request for you?
- Here, the agent reads the policy PDF and displays the response, even offering to initiate the carry-over request.
- Cost: 1 action = £0.08, cumulative cost = £0.16 + 1 happier employee.
If we estimate 100 employees, each asking an average of 5 HR-related questions per month over a year:
- Total actions = 100 × 5 × 12 = 6,000 actions
- Estimated cost: 6,000 × £0.08 = £480 per year (~£40 per month)
This demonstrates how an AI Agent can provide a cost-effective HR assistant.
This use case is a great example of why to use FlexCredits over Per Conversation credits.
- FlexCredits are ideal because employee queries vary month-to-month. The per-action model allows you to pay only for what is consumed, rather than a fixed conversation price.
- If employee usage grows, you can monitor consumption directly in Salesforce and adjust your FlexCredit purchases before costs escalate.
Use Case 2: Website Concierge (Chatbot Replacement) – illustrative
This use case explores an AI Agent helping potential customers with apartment recommendations, bookings, and additional activities. The agent has access to website content, available listings, and local events, enabling it to provide personalized guidance.
The prospective customer asks:
Customer: I am looking for an apartment that sleeps 4 with a pool in the south of France for June. What do you have available?
AI Agent: We have several listings available. Could you tell me a bit more about the specific area you would like in the south of France, length of stay and any activities you would like to do whilst on your trip? This will help me find you the best fit for you.
- The agent validates that listings exist matching the location, capacity, and pool requirements.
- It can check availability for additional activities and guide the conversation in a natural, exploratory tone.
Because interactions like this are more complex and variable, a fixed per-conversation charge is often more suitable than FlexCredits. For instance:
- A single conversation could include multiple passes to refine recommendations, conduct multiple availability checks and validate booking steps. When added together this can consume a lot of actions.
- Using the £1.60 per conversation model provides predictable costs while delivering an uncapped customer experience.
Capturing a qualified lead or completing a booking for £1.60 represents excellent value, particularly when compared with traditional manual handling by staff. If you use a chatbot on your website, you will no doubt, be familiar with the limitations and robotic tone it offers. Agentforce will brighten the interactions adding that human feel.
In Summary
From our experience, simply putting an AI Agent into the business to “see how it goes” rarely succeeds. Agentforce works slightly differently to other common LLM’s. It comes locked down by design and for good reason. You hold the key to its training. Sure, it does basic things out-of-the-box but still lacks the business context and use cases that bring the real magic to your business.
To maximise value, you need to identify where the efficiency gains or cost savings lie. Then deploy Agents to augment the business specific tasks, employee support and customer acquisition processes you have been building over the years.
We hope this article has gone some way to help you understand two key pricing options (FlexCredits & Per Conversation) as well as the Pro’s and Con’s of the different consumption models on offer to begin your AI journey on Salesforce.
What’s clear is that barrier to entry is extremely low, making it easier than ever to explore how AI agents can give you efficiency gains, cost savings or even a competitive edge in today’s fast moving AI landscape.
Salesforce is a platform we at Xenogenix have trusted for over a decade, and the introduction of Agentforce gives us a solution that we can confidently recommend.
We’re here to advise and support as always. Whether you want to explore use cases, understand consumption models, or simply chat about adopting Agentforce, we can help you navigate the process in a measured, informed way.
References:
https://www.salesforce.com/news/stories/new-agentforce-payment-options/
https://www.salesforce.com/agentforce/pricing/
Disclaimer:
Agentforce pricing is subject to change at any time, all pricing and costings listed above are correct as of 18/09/2025.