.png)
Frameworks, core principles and top case studies for SaaS pricing, learnt and refined over 28+ years of SaaS-monetization experience.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Inside Salesforce Agentforce 360 How Salesforce Prices Its Enterprise Grade AI Agents
Salesforce has spent less than two years turning Agentforce from a product with one memorable price into a portfolio with several ways to pay. At general availability on 29 October 2024, Agentforce Service Agent started at $2 per conversation. By 15 May 2025, Salesforce had added Flex Credits at $500 per 100,000 credits, with a standard action consuming 20 credits, or $0.10 at list price. A month later, it added employee-facing licences from $125 per user per month and Agentforce 1 Editions from $550. By 12 August 2026, buyers could choose among conversations, Flex Credits, per-user access, unmetered employee add-ons, bundled editions and several commitment structures.
The expansion is commercially important because Agentforce itself has moved well beyond a service chatbot. Salesforce describes Agentforce 360 as a platform spanning Sales, Service, Marketing, Commerce, Field Service, IT, Slack, data and cross-system workflows. By the first quarter of fiscal 2027, ended 30 April 2026, Salesforce reported $1.2 billion of Agentforce ARR and 3.8 billion Agentic Work Units delivered across Agentforce and Slack.
Monetizely's position is that Salesforce has largely solved Agentforce's packaging and billing-control problems, but its customer-facing pricing metric remains one generation behind the product. Agentforce 360 should make the Agentic Work Unit the primary enterprise meter, keep seats as an access and packaging tool, and move Flex Credits into the background as the rating currency. Until Salesforce makes that reset, customers can still end up optimising how an agent is designed to reduce billed actions rather than optimising what the agent actually accomplishes.
Salesforce has built a pricing portfolio because one conversation could no longer describe the work
The original $2-per-conversation price was easy to explain when Agentforce was introduced mainly through customer service. On 29 October 2024, Salesforce described agents that could resolve cases, qualify leads and take actions across business functions, but its published starting price was still tied to a conversation.
That unit became strained as the product moved into background work. Agentforce 2dx, announced in March 2025, could be triggered by data changes and operate without constant human oversight. By October 2025, Agentforce 360 covered cross-functional work from prospecting and campaigns to service, scheduling, quoting, IT support and commerce.
Salesforce's pricing history reads like a response to that widening scope.
| Date | Pricing or packaging move | What changed economically | Source |
|---|---|---|---|
| October 2024 | Agentforce Service Agent from $2 per conversation | One customer interaction was the dominant meter | Salesforce, 29 Oct 2024 |
| May 2025 | $500 per 100,000 Flex Credits; standard action = 20 credits, or $0.10 | Billing moved from a whole conversation to work performed inside it | Salesforce, 15 May 2025 |
| May 2025 | Flex Agreement | Customers could move investment between user licences and Flex Credits | Salesforce, 15 May 2025 |
| June 2025 | Agentforce add-ons from $125/user/month; Agentforce 1 Editions from $550/user/month | Salesforce reintroduced predictable seat economics for employee use | Salesforce, 17 Jun 2025 |
| October 2025 | Agentforce 360 became generally available | The product expanded into a multi-app, cross-workflow agent platform | Salesforce, 13 Oct 2025 |
| August 2026 | Flex Credits, conversations, $5 user licences, $125-$150 add-ons and $550+ Agentforce 1 Editions coexist | Salesforce now monetises the same agent platform through several commercial paths | Salesforce pricing page, accessed 12 Aug 2026 |
The progression matters more than any individual list price. Salesforce discovered that an autonomous agent is simultaneously software access, machine activity and completed work, and no single legacy SaaS meter neatly captures all three.
Enterprise customers also pushed Salesforce towards predictability. In its third-quarter fiscal 2026 earnings call on 3 December 2025, management described 16 Agentic Enterprise License Agreements signed in the quarter and roughly 100 in the pipeline, with those agreements described as multimillion-dollar deals. CRO Miguel Milano explained that large customers wanted relief from having to manage numerous consumption measures, while Marc Benioff acknowledged that customers had pressed Salesforce for more pricing flexibility after the original per-conversation approach.
That customer reaction explains why Salesforce now sells both consumption and unmetered employee access. It does not, however, resolve which meter best tracks value.
Engine CEO Elia Wallen captured what Salesforce got right in May 2025: “Salesforce’s new flexible pricing model allows us to use AI agents for different types of use cases far beyond customer service.”
Action pricing is flexible, but it makes workflow design part of the customer's bill
Flex Credits were a sensible correction to conversation pricing. A background agent updating a CRM record should not have to manufacture a "conversation" simply to fit the commercial model.
As of 12 August 2026, Salesforce charges $500 for 100,000 Flex Credits. A standard Agentforce action uses 20 credits, while an Agentforce Voice action uses 30. Salesforce defines actions broadly enough to include updating a record, summarising a case, answering a product question, executing a prompt or invoking a flow.
The arithmetic creates a useful break-even point. At list price, 20 standard actions cost $2, exactly matching one conversation.
| Annual customer workload | Average standard actions per conversation | Flex Credit cost | Conversation cost | Cheaper public meter |
|---|---|---|---|---|
| 100,000 conversations | 5 | $50,000 | $200,000 | Flex Credits |
| 100,000 conversations | 10 | $100,000 | $200,000 | Flex Credits |
| 100,000 conversations | 20 | $200,000 | $200,000 | Equal |
| 100,000 conversations | 30 | $300,000 | $200,000 | Conversations |
The table exposes the core pricing problem: under Flex Credits, two agents delivering the same customer outcome can cost different amounts simply because one workflow takes more platform actions.
Salesforce's own current examples make the distinction concrete. Its "Where is My Order?" flow uses two standard actions and costs $0.20 at published Flex rates. Its case-management example uses three actions for $0.30, while its field-service scheduling example uses five actions for $0.50. Salesforce notes that Data 360 and other consumption services can add further cost.
Customers therefore have an economic reason to inspect prompts, flows, retrieval calls and action boundaries. Some optimisation is healthy because it lowers compute and platform load. Yet a pricing metric becomes problematic when the buyer's finance team must understand agent choreography to know what a successful service interaction will cost.
Seat pricing produces the opposite effect. A $125-per-user monthly Agentforce add-on costs $1,500 annually for one employee, with unmetered employee Agentforce use under the published package. At the $0.10 standard-action rate, the same $1,500 buys 15,000 actions. Across 250 working days, the rough crossover is 60 actions per user per working day.
Salesforce has therefore created rational price arbitrage inside its own portfolio: low-activity users favour consumption; intensive employee users can favour a seat; long customer conversations can favour the conversation meter. Worse, Salesforce says Flex Credits and Conversations cannot be supported simultaneously in the same Salesforce org.
Commercial flexibility at the catalogue level does not automatically produce simplicity at the workload level.
Agentforce 360 has crossed the point where a human seat should anchor its economics
Monetizely's 5-Step Pricing Framework starts with goals and segmentation, defining what the company wants pricing to achieve and which customers or use cases differ meaningfully; moves to packaging, deciding what capabilities belong together; then selects the pricing metric, the unit on which customer spend grows; sets the rate, deciding how much to charge for each unit; and finishes with operationalisation, the metering, quoting, contracting, billing and usage controls needed to make the model work in practice. As discussed in Monetizing Agentic AI, the sequence matters because a mathematically attractive rate cannot rescue the wrong value metric. For Agentforce 360, the decisive issue is whether seats, conversations, actions or completed agent work best reflect the value created as autonomy rises.
The Agentic Monetization Spectrum, or AMS, sharpens that question for AI agents. It looks at three dimensions: zero-human ability, meaning how much useful work can be completed without human intervention; operational domain, meaning whether the agent handles one task, an end-to-end function or work spanning several functions; and the output/cost ratio, meaning whether useful output grows roughly with cost, improves by roughly 10:1 to 100:1, or can reach much steeper ratios. Greater autonomy, broader scope and a steeper output-to-cost curve weaken the logic of charging mainly for the human who happens to have access.
Agentforce 360 scores high enough that its economic centre has already shifted away from a conventional SaaS seat.
| AMS dimension | Agentforce 360 score | Evidence | Pricing implication |
|---|---|---|---|
| Zero-human ability | Large - 3/3 | Salesforce says agents can be triggered by data or APIs and act without human engagement; Engine, OpenTable and other customers report autonomous service use | Machine work, not employee access, should drive the main meter |
| Operational domain | Large - 3/3 | Agentforce 360 spans Sales, Service, Marketing, Commerce, Field Service, IT, Slack and cross-system actions | A single seat or conversation increasingly describes only one surface of the product |
| Output/cost ratio | Inflecting - 2/3 | Salesforce reported 3.8bn production AWUs by Q1 FY27; customer deployments report large automation and savings, but data, integration and action consumption remain material | Outcome-adjacent consumption is stronger than pure seats, but cost still needs guardrails |
| Total | 8/9 | High-autonomy, broad-domain agent platform | Primary pricing should migrate towards completed machine work |
The 8/9 score does not mean every Agentforce deployment produces exceptional returns. It means the product has the technical ability to perform work independently across enough functions that human headcount is no longer the cleanest denominator for price.
OpenTable's Senior VP of Global Customer Success, George Pokorny, reported during the March 2025 Agentforce 2dx announcement: “Three weeks in, it is handling 73% of all restaurant web queries.” Precina CEO John Oberg said: “Agentforce has completely transformed the way we support both patients and providers.”
The platform's economics increasingly come from what those agents finish, not merely how many OpenTable or Precina employees hold licences.
Salesforce's billing operations are stronger than the metric customers see
The strongest part of Salesforce's design appears at the final step of the 5-Step Framework. Consumption pricing requires finance-grade controls, and Salesforce has built several.
Digital Wallet provides near-real-time consumption visibility, threshold alerts and usage trends. Flex Credits support upfront purchase and pay-as-you-go, while Salesforce's published Pre-Commit structure bills usage monthly in arrears and includes a term-end true-up when consumption falls below the commitment. Over-entitlement Flex usage carries no published penalty rate; Salesforce says excess use is billed at the contracted rate. Unused Flex Credits, however, do not roll into later subscription terms.
Our scorecard therefore separates sophisticated commercial operations from the less settled choice of meter.
| 5-Step area | Grade | Monetizely assessment |
|---|---|---|
| Packaging | B+ | Strong entry points from $0 Foundations through consumption, employee add-ons and enterprise bundles, but overlapping choices increase buying complexity and Flex cannot coexist with Conversations in one org. |
| Pricing metric | B- | Flex Credits are better than conversation-only pricing, yet an action measures platform activity rather than whether the customer's job was completed. |
| Operationalisation | A- | Digital Wallet, alerts, PayGo, Pre-Commit, contracted overage treatment and licence-to-credit flexibility make consumption deployable at enterprise scale; non-rollover and true-up mechanics still transfer forecasting risk to buyers. |
What Salesforce gets right is the buying architecture. A company can begin at $0 with Foundations, put metered workers on a $5 Agentforce User Licence plus Flex Credits, use $125-$150 employee add-ons for heavier internal usage, or buy Agentforce 1 Editions from $550 per user per month. Current Agentforce 1 Editions also include 2.5 million Flex Credits per organisation per year, up from the one million announced with the package in June 2025.
What Salesforce gets wrong is allowing the rating unit to become the value story. Calling each technical action a value-bearing event does not make it an outcome. An agent may authenticate a customer, retrieve an order and provide a delivery date in two actions; another configuration may require four. The customer's value is still "customer received the correct order status".
Operational sophistication also creates pricing power. Salesforce's fiscal 2026 10-K says Agentforce can access live data through Data 360, follow Salesforce metadata and company policies, act through Salesforce applications and MuleSoft APIs, and operate with permissions, observability, governance and auditability. The same filing acknowledges that monetisation strategies for Agentforce and Data 360 remain relatively new and uncertain, and specifically warns that Salesforce has limited experience determining optimal prices for consumption contracts.
That disclosure is unusually important. Salesforce itself is telling investors that Agentforce's product maturity and pricing maturity are not yet the same thing.
Enterprise procurement strength reinforces the package. CrowdStrike Chief Business Officer Daniel Bernard said of Agentforce 360 for AWS: “The collaboration between Salesforce and AWS means we get proven AI capabilities with the procurement simplicity and security we need to scale.”
Those security, governance, data and purchasing advantages can justify a premium. They do not require the premium to be collected through an opaque action count.
Competitors are moving the charged event closer to completed value
The agent market provides a revealing comparison. Several B2B vendors have moved away from pricing AI solely through user access or raw interaction volume, although they differ sharply on what constitutes a billable event.
As of mid-2026, Intercom and HubSpot have gone furthest towards explicit outcome pricing. Microsoft remains closer to Salesforce's credit logic, while Zendesk has made automated resolution the organising unit for AI-agent consumption.
The comparison does not prove that Salesforce should copy Intercom's $0.99 outcome or HubSpot's $0.50 resolution. Salesforce's addressable work is much broader. A sales agent qualifying an opportunity, an IT agent resetting access and a commerce agent changing an order have different values.
What the table does show is the direction of travel. Service-focused products can define a successful resolution narrowly enough to bill for it. Broad enterprise agent platforms need a general unit of completed work that remains consistent across functions.
Salesforce now has one.
Salesforce should turn Agentic Work Units from an investor KPI into the primary commercial meter
In February 2026, Salesforce began reporting Agentic Work Units. By May, the company gave the term a precise definition: an AWU is a discrete task executed by an AI agent in production, with examples including resolving a customer case, updating a record and triggering an automated workflow. Salesforce reported 3.8 billion AWUs delivered to date in its first-quarter fiscal 2027 results, up 111% quarter on quarter, while Agentforce ARR reached $1.2 billion, up 205% year over year.
That combination creates the opening for the next pricing reset. Salesforce has already developed a customer-legible unit that sits above tokens and closer to real work than Flex Credits.
Monetizely's recommended architecture is therefore specific: make Agentic Work Units the primary enterprise Agentforce 360 meter; retain licences as the right to access particular applications and employee experiences; use Flex Credits internally to rate compute-heavy actions, voice and model services; and add outcome verification where a business result can be clearly confirmed.
The distinction matters. A customer should see "case resolved", "record updated", "appointment scheduled" or "workflow completed" on the commercial ledger. Salesforce can still translate the underlying model calls, actions, voice minutes and infrastructure into internal credit consumption to protect gross margin.
Such a reset would also align Salesforce's commercial model with its financial reporting. In the first quarter of fiscal 2027, Salesforce reorganised subscription revenue into two major categories, including Agentforce Apps and a Data 360, Headless Platform & Other category. Agentforce Apps explicitly includes Agentforce Apps Flex Credits, while the second category includes Data 360 and Platform Flex Credits.
Pricing has already become part of the product architecture rather than an add-on to it.
The 10-K makes the strategic tension clear. Most Salesforce subscription arrangements remain built around cloud-service access, typically with non-cancellable arrangements and revenue recognised over the contract term. At the same time, Salesforce says consumption pricing is becoming more prevalent and warns that usage below expectations can reduce revenue.
Agentforce therefore cannot simply abandon predictability. Large customers have told Salesforce they want it. The answer is not to retreat to seats as the primary value measure; it is to sell predictable commitments denominated in completed agent work.
For enterprise buyers and SaaS leaders designing their own agent pricing, four actions follow.
Run the AI business case in units of completed work before discussing licence counts. Track cases resolved, records changed, appointments completed, qualified opportunities or other concrete tasks, then map those workloads into vendor meters.
Give one executive owner responsibility for human and digital capacity together. Separate Service, Sales, Data and AI budgets make it too easy to add AI spending without testing whether corresponding operating capacity, revenue or service levels actually improve.
Treat meter migration as part of the product roadmap. Salesforce moved from conversation-only pricing to Flex Credits, per-user options and enterprise flat-fee structures within roughly 14 months. Buyers and vendors should expect today's AI meter to evolve as agents take on more work.
Make verified work the renewal currency. Salesforce now records billions of AWUs in production. Customers should use the same operational data to decide whether the next pound or dollar goes into more agent capacity, more human capacity or neither.
Salesforce deserves credit for moving faster on pricing than many large software vendors do. Conversation pricing got Agentforce into market. Flex Credits unlocked background work. Seat bundles answered enterprise demands for budget certainty. Digital Wallet and enterprise agreements made consumption governable.
But Agentforce 360 is no longer merely software that people use. It performs work. Salesforce's strongest pricing power will come when its invoice reflects that fact as clearly as its product does.
Assumptions
All modelled figures use published US-dollar list prices available on 12 August 2026, before negotiated volume discounts, taxes, implementation, base Salesforce licences, Data 360 consumption, partner services and other add-ons. Standard-action scenarios assume 20 Flex Credits per action and $500 per 100,000 credits. Employee break-even calculations use 250 working days per year. Conversation and Flex scenarios are compared economically even though Salesforce currently states that the two models cannot coexist within the same org.
Footnotes
Monetizing Agentic AI: https://www.amazon.com/Monetizing-Agentic-AI-Handbook-Transformation/dp/B0H7Z13VKJ/
Salesforce, Agentforce Pricing, accessed 12 August 2026: https://www.salesforce.com/in/agentforce/pricing/
Salesforce Investor Relations, Salesforce’s Agentforce Is Here: Trusted, Autonomous AI Agents to Scale Your Workforce, 29 October 2024: https://investor.salesforce.com/news/news-details/2024/Salesforces-Agentforce-Is-Here-Trusted-Autonomous-AI-Agents-to-Scale-Your-Workforce/default.aspx
Salesforce, Salesforce Introduces New Flexible Agentforce Pricing to Accelerate the Digital Labor Revolution, 15 May 2025: https://www.salesforce.com/in/news/press-releases/2025/05/15/agentforce-flexible-pricing-news/
Salesforce, Updated Product Packaging and Pricing Offer New AI Capabilities and More Ways to Scale AI Throughout Every Organization, 17 June 2025: https://www.salesforce.com/in/news/stories/pricing-update-2025/
Salesforce, Salesforce Announces the Agentic Enterprise, 13 October 2025: https://www.salesforce.com/news/press-releases/2025/10/13/agentic-enterprise-announcement/
Salesforce Investor Relations, Salesforce Launches Agentforce 2dx, 5 March 2025: https://investor.salesforce.com/news/news-details/2025/Salesforce-Launches-Agentforce-2dx-with-New-Capabilities-to-Embed-Proactive-Agentic-AI-into-Any-Workflow-Create-Multimodal-Experiences-and-Extend-Digital-Labor-Throughout-the-Enterprise/default.aspx
Salesforce Investor Relations, Salesforce Delivers Record First Quarter Fiscal 2027 Results, May 2026: https://investor.salesforce.com/news/news-details/2026/Salesforce-Delivers-Record-First-Quarter-Fiscal-2027-Results/default.aspx
Salesforce Q3 FY26 earnings-call transcript, 3 December 2025: https://www.investing.com/news/transcripts/earnings-call-transcript-salesforce-beats-q3-2026-forecasts-stock-climbs-93CH-4389747
Salesforce Investor Relations, Salesforce Delivers Record Third Quarter Fiscal 2026 Results Driven by Agentforce & Data 360, 3 December 2025: https://investor.salesforce.com/news/news-details/2025/Salesforce-Delivers-Record-Third-Quarter-Fiscal-2026-Results-Driven-by-Agentforce--Data-360/default.aspx
Salesforce, Annual Report on Form 10-K for fiscal year ended 31 January 2026, filed 2 March 2026: https://www.sec.gov/Archives/edgar/data/1108524/000110852426000060/crm-20260131.htm
Salesforce, Salesforce, AWS Deepen Collaboration to Launch Agentforce 360 for AWS, December 2025: https://www.salesforce.com/news/stories/agentforce-360-for-aws-announcement/
Intercom, Fin AI Agent outcomes, updated 26 June 2026: https://www.intercom.com/help/en/articles/8205718-fin-ai-agent-outcomes
HubSpot, HubSpot's Customer Agent and Prospecting Agent: Now you pay when the task is complete, 2026: https://www.hubspot.com/company-news/hubspots-customer-agent-and-prospecting-agent-now-you-pay-when-the-task-is-complete
Zendesk, Moving to automated resolutions from existing pricing plans, updated 4 June 2026: https://support.zendesk.com/hc/en-us/articles/6931689272090-Moving-to-automated-resolutions-from-existing-pricing-plans
Zendesk, Pricing Plans, accessed 2026: https://www.zendesk.com/pricing/
Microsoft, Copilot Studio Licensing Guidance, accessed 2026: https://www.microsoft.com/licensing/guidance/Microsoft-Copilot-Studio