TL;DR: Salesforce Agentforce costs range from $143K to over $1.65M in year one depending on deployment scope. The published pricing (starting at $2 per conversation or $500 per 100,000 Flex Credits) does not include Data Cloud (required, from $108K/year), implementation ($25K to $850K), or knowledge base setup. This guide breaks down all four pricing models, the hidden costs, and how to model total spend before signing.
What Salesforce Agentforce actually costs
Salesforce publishes Agentforce starting prices on their pricing page. Those numbers are technically accurate and completely misleading if you use them to build a budget. The starting prices cover the agent runtime only. The rest of what you need to actually deploy Agentforce (Data Cloud, knowledge base structure, agent configuration, ongoing tuning) sits outside those numbers.
For an enterprise deployment covering 3 to 5 use cases, the honest year-one total is $310,000 to $630,000. For larger deployments across service, sales, and marketing, the number crosses $1 million. If you are still deciding whether Agentforce is the right platform at all, our breakdown of Agentforce, HubSpot Breeze, and the autonomous revenue stack compares the options before you get to pricing.
The three current Agentforce pricing models
Salesforce runs three Agentforce pricing models simultaneously in 2026, plus a fourth outcome-based model launched in July for Help Agent. The choice among them is not obvious, and picking wrong locks you into worse economics for years.
1. Per-conversation pricing ($2 per conversation)
Launched at Dreamforce 2024 as the original Agentforce model. Still available, but now scoped to customer-facing agents under a fixed model.
Best for: Small customer service teams testing Agentforce with predictable, low volume.
Watch out for: What counts as a “conversation” is not always what buyers expect. A single customer query triggering 8 backend processes can bill as one conversation, or in some configurations, more.
2. Flex Credits ($500 per 100,000 credits, ~$0.10 per action)
Introduced May 2025 to solve the conversation-counting problem. Buyers purchase credits in bulk and spend them at the action level. Each agent action (retrieving data, running a flow, generating a response) costs approximately $0.10 in credits.
A typical customer service conversation involves 8 to 15 actions, so the effective cost lands between $0.80 and $1.50 per conversation, which is cheaper than per-conversation pricing at high volume.
Best for: Multi-team deployments running customer, employee, and voice use cases together.
Watch out for: Credit consumption is nearly impossible to predict before you have been running agents in production for 2 to 3 months. Overestimating means locked-up cash. Underestimating means unexpected overage charges. Where multiple agents run across marketing, sales, and service, consumption is also a function of how well those agents are coordinated, which is a RevOps AI orchestration problem before it is a budgeting one.
3. Per-user licensing ($125 to $550+ per user per month)
Added in 2025 as an alternative to consumption-based billing. Agentforce add-ons start at $125 to $150 per user per month. Agentforce 1 Editions (the full enterprise version) start at $550+ per user per month.
Best for: Employee-facing deployments at scale, where you know exactly how many people will use Agentforce and want predictable monthly costs.
Watch out for: The math only works if utilization is high. A $150/month license for a user who runs 5 agent interactions per week is dramatically more expensive than per-action pricing would be.
What is Pay-Per-Resolution pricing?
Pay-Per-Resolution is Salesforce’s newest pricing model, launched in July 2026 for Agentforce Help Agent. You are charged only when the agent autonomously resolves a customer’s issue. If the customer asks for a human or walks away unhappy, there is no charge.
This is genuinely different from the earlier models. Per-conversation and per-action billing charged for activity. Pay-per-resolution charges for outcomes.
For customer service teams, this reframes the ROI conversation entirely. Instead of “how much does the agent cost us per month,” the question becomes “what is our cost per resolved ticket.” That is a number CFOs can actually model against internal support costs, and it is the kind of metric a mature revenue operations function should already be tracking.
Pay-per-resolution is currently limited to Help Agent. Salesforce has not yet extended it to sales, marketing, or IT-facing agents.
Data Cloud is required, and it is not cheap
The single biggest hidden cost in every Agentforce deployment is Data Cloud. Agentforce runs on Data Cloud. That is not optional. It is architectural. Our Salesforce consulting practice scopes Data Cloud as part of the architecture from day one for exactly this reason.
Minimum Data Cloud commitment for Agentforce customers runs approximately $108,000 per year. Enterprise deployments with high data volumes routinely spend $150,000 to $300,000+ per year on Data Cloud alone.
This is the line item that most 2024 Agentforce projections missed. The $2 per conversation math looked friendly, but customers who did the math on Data Cloud usage discovered that the platform costs added up faster than the agent conversation costs did.
For any enterprise modeling an Agentforce budget, start with the Data Cloud number and build up from there. Not the other way around.
Implementation costs by scope
Salesforce sells Agentforce as easy to deploy. In simple orgs with clean data models, that is roughly accurate. In enterprise orgs with 5+ years of Salesforce customization, deployment is a real project. That accumulated customization is technical debt, and it is the variable that moves implementation cost more than anything else on this list.
Realistic implementation costs by scope:
Simple deployment (1 to 2 use cases, clean Salesforce org)
- Agent configuration: $15,000 to $50,000 per use case
- Knowledge base setup: $10,000 to $30,000
- Total: $25,000 to $80,000
Mid-market deployment (3 to 5 use cases, some data cleanup required)
- Agent configuration: $60,000 to $150,000 combined
- Knowledge base setup and expansion: $20,000 to $50,000
- Data cleanup and readiness work: $30,000 to $80,000
- Total: $110,000 to $280,000
Enterprise deployment (5+ use cases, complex Salesforce org)
- Agent configuration: $200,000 to $400,000
- Knowledge base setup: $50,000 to $150,000
- Data cleanup and readiness work: $80,000 to $200,000
- Change management and training: $30,000 to $100,000
- Total: $360,000 to $850,000
The data cleanup line is the one buyers most often try to cut, and it is the one that decides whether the agent works. Dirty records do not produce a broken agent, they produce a confident wrong one. Our CRM optimization guide covers what “AI ready” data actually means, and why most agent pilots stall within three months when that work gets skipped.
None of these numbers include ongoing tuning, which typically runs 15 to 25 percent of implementation cost annually.
Total year-one cost by deployment size
Bringing it all together, this is what year-one Agentforce spend actually looks like across three deployment sizes. Numbers are based on published Salesforce pricing plus implementation ranges from partner ecosystem data.
These numbers are directional. Actual costs depend heavily on your existing Salesforce org state, data quality, and how tightly you scope the initial deployment. At the enterprise end of this table, the architecture decisions matter more than the pricing model, which is the subject of our enterprise AI architecture and GTM blueprint. You can also see modelled outcomes from comparable deployments in our case study portal.
Which pricing model should you choose?
The choice depends on volume, predictability, and use case mix.
Choose per-conversation if:
- You are running one customer-facing agent
- Volumes are under 10,000 conversations per month
- You want a simple line item in your budget
Choose Flex Credits if:
- You are running multiple agents across use cases
- You want granular action-level cost tracking
- You have 3+ months of usage data to model consumption
Choose per-user licensing if:
- You are deploying employee-facing agents at scale
- User utilization will be high (5+ interactions per user per day)
- You want fixed, predictable monthly costs
Choose pay-per-resolution if:
- You are deploying Help Agent for customer service
- You want to tie cost directly to business outcomes
- Your existing self-service resolution rate is already tracked
Most enterprise buyers negotiate hybrid contracts that combine models: Flex Credits for some agents, per-user for others, with conversion rights if usage patterns change. If you are weighing Agentforce against the HubSpot side of the market, our RevOps by platform comparison sets out where each stack earns its cost.
How to negotiate an Agentforce contract
Three levers matter most in Agentforce negotiations:
1. Ask for conversion rights between pricing models
Do not lock into one model for the whole term. Ask for the right to switch to a different model at annual renewal, or to convert if usage crosses a threshold.
2. Negotiate Data Cloud pricing separately, or bundle it up front
Data Cloud is where Salesforce has the most flexibility on discounts. Bundling Data Cloud into your master agreement typically produces better economics than adding it as a separate contract.
3. Cap overage exposure with pre-commit plus true-up
Pre-commit pricing is often the best of both worlds: discounted rates for committed baseline usage, with monthly true-ups for actual consumption. Avoid pure pay-as-you-go for enterprise deployments; overage exposure can be significant.
For any deployment over $250,000 year one, invest in an experienced Salesforce partner for the contract review. The negotiation ROI typically covers the partner fee 3x to 5x over. We compared the field in Best Salesforce Agentforce Implementation Consultants (2026), including which firms audit your data before they deploy and which do not.
One clause worth adding while you have leverage: governance. Agent audit trails, human approval gates, and data residency terms are far easier to negotiate before signature than after. Less than 10 percent of enterprises are ready for agentic AI at scale, and EU AI Act enforcement has been live since 2 August 2026.
Ready to model your Agentforce budget?
Before you sign an Agentforce contract, run a Salesforce readiness audit. It reveals what your specific org needs before deployment, which pricing model fits your use cases, and which hidden costs will hit you if you move forward without preparation.
Book a 30-minute discovery call at explore.mountainise.com. No sales pitch. Just a look at your current Salesforce org and what an honest Agentforce deployment would cost you.
Frequently Asked Questions
Agentforce has four active pricing models: $2 per conversation, Flex Credits at $500 per 100,000 credits (~$0.10 per action), per-user licenses from $125 to $550+ per user per month, or Pay-Per-Resolution for Help Agent. For a typical enterprise deployment, all-in monthly costs land between $12,000 and $80,000+ depending on scope.
Yes. Data Cloud is required for Agentforce. Minimum annual Data Cloud commitment for Agentforce customers is approximately $108,000, with enterprise deployments typically spending $150,000 to $300,000+ annually. Our CRM implementation team scopes this as part of the platform architecture rather than as an add-on.
Per-conversation charges $2 per completed conversation regardless of complexity. Flex Credits charge approximately $0.10 per agent action, so a typical 8 to 15 action conversation costs $0.80 to $1.50. Flex Credits are typically cheaper at high volume but harder to predict in advance.
Pay-per-resolution is Salesforce’s July 2026 pricing model for Agentforce Help Agent. You are charged only when the agent autonomously resolves a customer issue. No charge if the customer asks for a human or walks away.
Yes. Salesforce Foundations gives Enterprise Edition customers no-cost access to a limited Agentforce tier as a permanent option, not a trial. Any deployment beyond a single agent typically requires moving to paid tiers.
Enterprise Edition is the minimum for most Agentforce configurations. Agentforce 1 Editions (starting at $550+ per user per month) unlock the full agentic capability set.
Simple deployments (1 to 2 use cases): $25,000 to $80,000. Mid-market (3 to 5 use cases): $110,000 to $280,000. Enterprise (5+ use cases in complex orgs): $360,000 to $850,000. Ongoing tuning adds 15 to 25 percent annually.
Vendor pitch: 4 to 6 weeks. Real enterprise timeline: 22 to 44 weeks depending on org complexity. The gap is almost always pre-deployment data work, which is why we recommend a structured readiness audit first. See our comparison of RevOps AI readiness audit firms for what that audit should actually cover.