SaaS Critique

Home / Reviews / AI for sales and support

Salesforce Agentforce review (2026): The obvious AI agent platform for Salesforce shops, with pricing that still takes work to understand.

Agentforce is Salesforce's platform for building and running AI agents that answer customers, work cases, qualify leads and assist employees inside Salesforce. It is deeply tied to your CRM data and permissions and now has flexible credit, conversation and per-user pricing, but it makes most sense for existing Salesforce customers, and costs remain hard to forecast.

GAdd SaaS Critique as a preferred source on Google

SaaS Critique is free to read. Some vendors pay for sponsored placements in our best lists, which helps fund the site. Payment never changes a score or a verdict, and we cover products whether or not they pay us. How we score.

7.6/10Try firstThe obvious AI agent platform for Salesforce shops, with pricing that still takes work to understand.

Key takeaways

  • Best for: Salesforce customers deploying AI agents across sales and service.
  • Price: Free to start; Flex Credits $500 per 100k.
  • Buy it if: You already run Salesforce and want AI agents that resolve service cases, assist reps or answer customers using your own CRM data and permissions.
  • Skip it if: You do not use Salesforce, want a simple fixed-price support chatbot, or lack the admin or partner capacity for a scoped rollout.

AI for sales and supportSalesforce Agentforce at a glance

Pros

  • Agents act on live CRM data and respect existing Salesforce security, sharing and governance
  • Covers customer-facing service agents, sales agents, employee assistants, voice and Slack
  • Flex Credits cut the entry cost to about $0.10 per standard action, far below the original $2 per conversation
  • Agent Script and Agentforce Builder give admins more predictable control over agent behavior
  • Free Salesforce Foundations tier lets you start building before buying usage

Cons

  • Best value only if you already run Sales Cloud, Service Cloud or other Salesforce clouds
  • Three pricing models (Flex Credits, Conversations, per-user add-ons) make total cost hard to forecast
  • Per-user add-ons are expensive at $150 to $175 a user a month, and Max editions start at $550
  • Implementation usually needs clean data, careful scoping and often a Salesforce partner
  • Early deployments were buggy; practitioners say documentation for advanced setups still lags

What Agentforce is

Agentforce is Salesforce's platform for autonomous AI agents. Announced in September 2024, it lets companies build agents that understand a request, decide which steps to take and carry them out using Salesforce data, flows and integrations. Typical agents answer customer questions on a website or phone line, resolve service cases, qualify and nurture leads, and help employees find information or update records.

It has grown into a family of products: Agentforce Builder and Agent Script for creating agents, Agentforce Voice for phone conversations, Slack-based agents, the employee-facing Agentforce Coworker assistant and Agentforce Vibes. It sits on top of Salesforce's CRM, Data 360 (formerly Data Cloud) and its trust and governance layer.

Who Agentforce is for

Agentforce is built first for organizations already using Salesforce, especially Service Cloud and Sales Cloud. Because agents act on CRM records with existing permissions, the value is highest when customer, case and order data already live in Salesforce.

Large enterprises with many service interactions and internal processes get the most from it, but Flex Credits have made pilots realistic for smaller organizations and nonprofits too. Companies that do not use Salesforce, or that just want a fixed-price website chatbot, will usually find simpler and cheaper tools elsewhere.

Key features

Customer-facing agents handle requests on web, messaging and voice. Salesforce's own example walks through an agent authenticating a customer, retrieving their latest order from an order management system and giving a real-time status update. Agentforce Voice extends this to phone calls, starting each call with Salesforce context and handing off to humans when needed; Salesforce told Salesforce Ben in November 2026 that Voice had passed 10 million annualized calls.

Employee-facing agents work inside Sales, Service and Field Service to summarize cases, draft responses, update records and surface insights. Agentforce Coworker, now sold as a standalone per-user product, is an AI teammate that searches across connected enterprise data while applying Salesforce security and access policies. Independent reporting notes that Coworker is powered by Anthropic's Claude.

Building tools have matured. Agentforce Builder and Prompt Builder are low-code, and Agent Script gives admins a more structured way to write agent instructions, which practitioners say makes behavior more consistent than early free-text instructions. Agents can call Flows, Apex and external systems through MuleSoft, and Data 360 grounds responses in unified customer data.

Operations and governance features include Digital Wallet for real-time credit consumption tracking, testing and deployment across sandboxes, and Salesforce's trust layer for data masking and auditing.

Pricing explained

Agentforce pricing has changed several times since launch. The current page lets you start free with Salesforce Foundations, then pay by consumption through Flex Credits or Conversations, or by user through add-ons, the Agentforce User License, Coworker or Max editions.

Flex Credits are the easiest way to begin: about $0.10 per standard action makes a narrow pilot inexpensive. As usage grows, costs depend on how many actions each conversation triggers, so a complex agent can cost several times more per interaction than a simple one.

For heavy internal use, per-user add-ons remove metering but at a high seat price. Most enterprises negotiate a mix, and some sign unlimited agreements. Budget time to model costs with your Salesforce account team before committing.

Ease of use and setup

Admins familiar with Salesforce can build a basic agent with the low-code tools, but production deployments need clear scope, decent data and testing. Implementation experts interviewed by Salesforce Ben said early 2025 versions were buggy, lost context and behaved differently across environments, and that stability and documentation improved noticeably by the end of 2025.

Their advice is to start with a narrow, controllable use case with clear return on investment, then expand. You do not need a perfect org, but pilots often expose data and process gaps that must be fixed. Many customers use Salesforce partners for implementation.

Integrations

Agentforce works natively with Salesforce clouds, including Sales, Service, Marketing and Commerce Cloud, Field Service, Industries clouds, Slack, Tableau and Data 360. MuleSoft and APIs connect agents to external systems such as order management or ERP.

Because agents use Salesforce's metadata, security and sharing rules, integrations with your existing automation, Flows and Apex are a key strength. Non-Salesforce data needs to be connected through Data 360, MuleSoft or other integration tools first.

Support and reliability

Support follows your Salesforce support plan, and Salesforce has expanded Trailhead courses, Agentblazer learning paths and best-practice guides. Practitioners still say finer configuration topics, such as keeping context when deploying across sandboxes, are thinly documented and often left to partners.

Agentforce runs on Salesforce's enterprise platform and trust layer. Customer stories are mixed: OpenTable and Adecco report substantial deployments, with Salesforce reporting OpenTable deflects an estimated 70% of tier-one restaurant support cases, while Saks Fifth Avenue, an early showcase customer, has used a competing vendor for its voice assistant while remaining a Salesforce customer.

What users like and dislike

Buyers and implementers value Agentforce's direct access to CRM data, its security model, the breadth of agent types and the lower entry cost of Flex Credits. Salesforce reported 18,500 Agentforce deals in 2025, with Agentforce and Data 360 together reaching $1.4 billion in annual recurring revenue.

The main complaints are pricing complexity and uncertainty, early bugs and context loss, documentation gaps for advanced scenarios, and the effort needed to prepare data. Some customers bought Foundations without understanding credit consumption, and buyers increasingly want proof from production deployments rather than demos.

Our verdict

If your business runs on Salesforce, Agentforce is the most natural way to add AI agents, because it works directly with your records, flows and permissions. Flex Credits make it cheap to prove value on a narrow use case first.

If you do not use Salesforce, or want a simple support bot with a predictable monthly bill, look at dedicated AI support tools instead. Either way, pilot carefully, measure deflection and resolution, and model costs before scaling.

Salesforce Agentforce scores 7.6 out of 10 overall. Its strongest criterion is room to grow (9.5/10) and its weakest is value for money (6/10).

Salesforce Agentforce pricing

PlanPriceWhat you get
Salesforce Foundations$0Start building with Agentforce Builder, Prompt Builder, Agent Script, Coworker and Vibes
Flex Credits$500 per 100,000 creditsPay per action: standard actions use 20 credits (about $0.10), voice actions 30 credits; works for customer- and employee-facing agents, Voice and Slackbot
ConversationsCustom quoteFlat price per conversation for customer-facing agents
Agentforce add-ons$150 per user/mo (Industries $175)Unmetered employee Agentforce usage in Sales, Service and Field Service, plus AI analytics and Prompt Builder
Agentforce User License$5 per user/mo plus Flex CreditsCompany-wide agent access with limited CRM objects and metered usage
Agentforce Coworker$50 per user/moStandalone AI teammate with all Coworker searches included and 30 MB indexing per user a month
Max EditionFrom $550 per user/moSales, Service or Industries edition with Agentforce add-on and 2.75M Flex Credits per org a year

Salesforce offers three ways to pay. Flex Credits cost $500 per 100,000 credits, and each agent action, such as updating a record, summarizing a case or answering a product question, draws 20 credits, or 30 for Agentforce Voice. Salesforce's own example of 20 two-action requests a day comes to 24,000 credits, or about $120 a month.

Conversation pricing gives a flat price per customer-facing conversation and is quoted by sales. Salesforce originally launched Agentforce at $2 per conversation in 2024; independent coverage credits the move to Flex Credits with making pilots affordable for nonprofits and smaller businesses.

Per-user options suit internal use: Agentforce add-ons at $150 a user a month ($175 for Industries clouds) give unmetered employee usage, Max editions start at $550 a user a month with 2.75 million Flex Credits per org a year, the Agentforce User License costs $5 a user a month plus credits, and Agentforce Coworker is $50 a user a month.

Credits can be bought upfront (Pre-Purchase), committed over the term (Pre-Commit) or paid as you go (PayGo), with usage tracked in Digital Wallet. Salesforce says the page is informational and subject to change. Practitioners in 2026 still describe working out the real cost of a given use case as a frequent question that takes time with Salesforce sales teams.

Prices in US dollars, checked on Nov 21, 2026. Confirm on Salesforce Agentforce's own site before you buy.

Buy it ifYou already run Salesforce and want AI agents that resolve service cases, assist reps or answer customers using your own CRM data and permissions.

How we scored Salesforce Agentforce

CriterionScoreWhy
Value for money (25%)6Flex Credits make small pilots affordable, but per-user add-ons and Max editions are costly and total spend is hard to predict.
Features (25%)9Agents, builder tools, voice, Slack, Coworker, Data 360 grounding and deep CRM actions make it one of the most complete agent platforms.
Ease of use (20%)6Low-code tools have improved, yet setup still depends on data quality, Salesforce admin skills and often partner help.
Room to grow (15%)9.5Built for enterprises, with pre-commit and pre-purchase models, org-wide credit pools and unlimited agreements.
Support and trust (15%)8Runs on Salesforce's trust, security and governance layer, though the product's early bugs and mixed customer stories call for a careful pilot.

Alternatives to Salesforce Agentforce

  • Zendesk AI agents: Choose it if your support team runs on Zendesk and wants AI resolution tied to its help desk.
  • Intercom Fin: Choose it for a fast-to-deploy customer service AI agent priced per resolution.
  • Microsoft Copilot Studio: Choose it if your organization runs on Microsoft 365 and Dynamics 365.
  • HubSpot Breeze: Choose it for AI agents built into HubSpot's CRM for small and mid-sized businesses.
  • Sierra: Choose it for a dedicated, enterprise conversational AI agent platform independent of your CRM.

Salesforce Agentforce FAQs

How much does Agentforce cost?

You can start free with Salesforce Foundations. Flex Credits cost $500 per 100,000 credits, with a standard action using 20 credits (about $0.10). Per-user options include $150 a user a month add-ons, a $5 a user a month User License plus credits, Coworker at $50 a user a month and Max editions from $550 a user a month. Conversation pricing is quoted.

What are Flex Credits?

Flex Credits are Salesforce's unit for Agentforce usage. Each action an agent takes, such as updating a record or answering a question, uses 20 credits, or 30 for Agentforce Voice. Credits can be used across teams and use cases and tracked in Digital Wallet.

Is Agentforce still $2 per conversation?

That was the original 2024 pricing. Salesforce now offers Flex Credits and quoted Conversation pricing, and independent coverage reports that per-action costs can be far lower than $2.

Do I need Salesforce to use Agentforce?

In practice, yes. Agentforce is built on the Salesforce platform and works best with data in Salesforce clouds and Data 360.

What happens if I run out of Flex Credits?

Salesforce's FAQ covers overage handling for each buying model; you can buy more through Pre-Purchase, Pre-Commit or PayGo models. Check terms with your account executive.

Is Agentforce ready for production?

Many companies run it in production, and implementers say stability improved greatly after early 2025. Start with a narrow use case, test thoroughly and expand once results are proven.

Sources

Checked on 2026-11-21. This is a desk review based on the sources below and published user feedback, not a long-term hands-on test.

Why you can trust SaaS Critique

Every product on SaaS Critique is scored on the same five criteria with the same weights, and each review ends with who should not buy it. We work from vendors' public pricing pages, plan tables, documentation and patterns in published user feedback. Unless a review says otherwise, it is a desk review rather than a long-term hands-on test.

Some best lists include sponsored placements. A sponsored placement is a position in a list. It never changes a product's score, its verdict or what we say about its weaknesses.

Our methodology
  • Value for money (25%). What you pay against what you get, including the plan most teams end up on.
  • Features (25%). How much of the job the product covers without add-ons.
  • Ease of use (20%). How quickly a new person can do useful work in it.
  • Room to grow (15%). Whether it still works when the team and the workload double.
  • Support and trust (15%). Support quality, pricing stability and how the vendor treats customers.

Overall score of 8.0 or more: Buy. 6.5 to 7.9: Try first. Below 6.5: Skip. Read the full method.

Compare Salesforce Agentforce with

Nathan Brooks

About the author

Nathan Brooks

Editor, SaaS Critique

Nathan edits SaaS Critique and writes our best lists and reviews of business software for small and mid-sized teams. He checks every list against vendors' current pricing pages and holds each product to the same five-part scoring method, so every verdict can be compared with the next.