Agentforce Pricing 2026: What It Costs

As of September 2026, Agentforce pricing comes in three main shapes: Flex Credits at $500 per 100,000 credits (about $0.10 per standard action), $2 per conversation for customer-facing agents, or per-user add-ons from $125 per user per month. The license is rarely the hardest part. Getting your data into a state an agent can use usually is, so price one job first and buy for that.

How much does Agentforce cost?

Salesforce publishes its Agentforce pricing, and the options below are what its pricing page listed when we checked it on 30 September 2026. Pricing like this changes, so treat it as a snapshot, not a quote.

Option	List price (as of September 2026)	Notes
Flex Credits	$500 per 100,000 credits	A standard action uses 20 credits (about $0.10); a Voice action uses 30. Pre-purchase, pre-commit or pay-as-you-go.
Conversations	$2 per conversation	Customer-facing agents only.
Agentforce add-ons (Sales, Service, Field Service)	$125 per user per month	Industries add-ons are $150.
Agentforce 1 Editions	From $550 per user per month	Includes 2.5 million Flex Credits per org per year.
Agentforce User License	$5 per user per month	Plus Flex Credits.
Salesforce Foundations	Free	Check what it includes for your edition.

Those are list prices. What you actually pay depends on your contract, your edition and what you already own, which is why the first question on any Agentforce call should be "what are we already licensed for?"

How do Flex Credits work in practice?

Flex Credits are metered by the action, not by the user. The arithmetic from the published rates is simple enough to do on a napkin:

$500 buys 100,000 credits, which covers 5,000 standard actions at 20 credits each.
A job that runs 10,000 standard actions a month uses 200,000 credits, or $1,000 a month at list.
The same $500 covers about 3,333 Voice actions at 30 credits each.
The 2.5 million credits bundled with an Agentforce 1 Edition cover 125,000 standard actions a year per org.

The unknown is not the rate, it's the volume. Before you choose a model, ask whoever designs the agent how many actions one task will take, and multiply by how many times a month that task really happens. If you can't answer that yet, you're not ready to pre-commit to anything.

The rough logic most buyers land on: per-user add-ons when a defined team will use agents all day, Flex Credits when volume is uncertain or spread thin, and per-conversation pricing only for agents that talk to customers.

What does an Agentforce implementation need first?

The prerequisite is data, specifically Data 360, which Salesforce called Data Cloud until it renamed it at Dreamforce in October 2025. According to Salesforce's own setup documentation, Data 360 must be provisioned and enabled for Agentforce, and the Einstein Trust Layer and Data Library depend on it. Enterprise Edition and above can provision it at no cost.

"Enabled" is not the same as "implemented." Switching Data 360 on satisfies the setup checklist. It does not connect your ERP, remove your duplicate accounts or decide which system holds the true customer address. That work is where Agentforce implementation budgets tend to go.

Gartner has put a number on the risk. In February 2025 it predicted that through 2026, organizations will abandon 60% of AI projects that lack AI-ready data, and it found that 63% of organizations either lack, or are unsure they have, the right data-management practices for AI.

A short readiness check before you buy:

Which records will the agent read, and which system is the source of truth for each?
Are those records deduplicated and current, or does your team keep the real numbers in a spreadsheet?
Which integrations does the agent need, and do they exist today?
Who is allowed to see what, and does the agent inherit those rules?

What does this mean for healthcare and life sciences?

Clinicians are not waiting for permission. The American Medical Association's 2026 physician survey found that 81% of physicians use AI, against 38% in 2023, and 72% on a stricter measure of at least one named use case. The same survey lists privacy among physicians' top concerns.

For a healthcare or life sciences organization, that makes the data question sharper, not softer. Whatever your compliance team requires of patient or trial data applies to every record an agent can read. Settle permissions and data boundaries before the pilot, not after the first surprised email from compliance. If the agent will talk to patients or members directly, it falls under the customer-facing conversation pricing, so model that volume separately.

Why start with one Agentforce job?

Plenty of companies are buying. Salesforce reported Agentforce annual recurring revenue above $1.5 billion in its second quarter of fiscal 2027 (August 2026), up more than 240% year on year. Buying is not the same as getting value, though. Gartner predicted in June 2025 that over 40% of agentic AI projects will be cancelled by the end of 2027 because of rising costs, unclear value or weak risk controls.

All three of those causes shrink when the scope is one job. A good first job has:

Clean, known inputs that already live in Salesforce or can be connected cheaply.
An output you can count: cases closed, hours saved, follow-ups sent on time.
A volume you can estimate, so the Flex Credit math is real.
A team that was shown why it helps them, not just told it's coming.

Prove it there, measure it for a few months, then widen. It's slower than a company-wide launch and much cheaper than a cancelled one.

How should you choose an Agentforce pricing model?

In order:

List what you already license. Foundations, existing editions and any bundled credits change the math.
Pick one job and estimate its monthly action or conversation volume.
Check the data that job needs against the readiness questions above. Fix that first.
Price the job under two models and compare at your real volume, not the demo's.
Buy the smallest commitment that covers the pilot. Pay-as-you-go exists for this.

If you'd like a second set of eyes, our Salesforce and AI work is delivered by MSquare Technology, a certified Salesforce partner. To be plain about the money: MSquare pays Salter Growth if an engagement goes ahead, and you pay us nothing. You can book a 30-minute call and we'll start with what you already own.

Agentforce pricing, and what to fix before you pay it