Salesforce Agentforce: Features, Pricing, Models & Alternatives (2026)

Last researched: September 26, 2026. Flex Credit, conversation, and per-user prices below were checked against Salesforce's own pricing page (https://www.salesforce.com/agentforce/pricing/) and should be reconfirmed there, since Agentforce's consumption-based pricing has changed more than once since launch. Agentforce is Salesforce, Inc.'s AI agent platform, running on what Salesforce now calls the Agentforce 360 Platform (renamed from the Einstein 1 Platform at Dreamforce 2025).

What Is Salesforce Agentforce?

Agentforce lets organizations build and deploy autonomous AI agents — for customer service, sales, and internal employee workflows — that act directly on live Salesforce CRM and Data Cloud records rather than a static knowledge base. Its core differentiator is the Atlas Reasoning Engine, an orchestration layer that plans, acts, and observes in a loop until a goal is met, drawing on multiple specialized models rather than one general-purpose LLM. Salesforce reported roughly 29,000 Agentforce deals and about $800M in related ARR by Q4 of its FY26, though independent analysis suggests adoption still sits at a low single-digit percentage of Salesforce's overall customer base.

Key Capabilities

• Atlas Reasoning Engine orchestrates multiple LLMs, large action models (LAMs), and a retrieval-augmented-generation module, reportedly drawing on 8-12 specialized models per query.
• Agentforce Builder and Prompt Builder provide low-code tooling to design, test, and deploy custom agents.
• Agent Script offers a scripting layer for more deterministic, developer-controlled agent behavior.
• Prebuilt agent templates exist for Service, Sales, and other Salesforce Clouds, alongside Agentforce Voice for telephony-based agents.
• Agentforce Coworker and Agentforce Vibes target employee-facing productivity use cases rather than only customer-facing ones.
• Digital Wallet and Slackbot integrations extend agents into payment and collaboration workflows.

Main Features

| Agentforce Coworker / Vibes | Employee-facing AI assistance and productivity tooling | | Digital Wallet / Slackbot | Payment and Slack-based agent integrations |

Use Cases

• Deflecting and resolving inbound customer service cases autonomously
• Qualifying and following up with sales leads without human intervention for routine steps
• Answering employee questions grounded in internal CRM/Data Cloud records
• Automating multi-step workflows that previously required custom Apex/Flow development
• Voice-based customer support agents integrated into existing telephony

Supported Models / Technologies

The Atlas Reasoning Engine is described by Salesforce as model-agnostic, selecting from roughly 8 to 12 specialized large language and large action models per query rather than depending on one model; which specific third-party model providers (e.g., OpenAI, Anthropic, Google) are active in a given deployment is not fully and publicly itemized by Salesforce as of this writing, and should be confirmed through Salesforce's own trust/model documentation for a given org.

Platforms

• Web (Salesforce Lightning/CRM)
• Mobile (Salesforce mobile app)
• Slack
• Voice/telephony (Agentforce Voice)
• API

Integrations

Salesforce Data Cloud (Data 360), MuleSoft, Tableau, Slack, and Salesforce's own Sales Cloud, Service Cloud (Agentforce Service), Marketing Cloud, and Commerce Cloud products, plus third-party systems via API and MuleSoft connectors.

Pricing

Most advanced, production-scale deployments also require Salesforce Data Cloud (Data 360), which independent cost analyses cite as pushing realistic first-year costs into the $150,000-$600,000 range for mid-market teams — well above the base Agentforce line items above.

Advantages

• Multi-model orchestration (Atlas) avoids hard dependency on a single LLM vendor
• Deep native grounding in existing Salesforce CRM and Data Cloud records
• Several distinct pricing mechanisms (credits, per-conversation, per-user) let teams match billing to their usage pattern
• Free Foundations tier gives existing Enterprise+ customers a genuine no-cost starting point

Limitations

• Real-world total cost of ownership is hard to estimate up front because of multiple overlapping billing mechanisms plus required Data Cloud spend
• Independent analysis suggests adoption (around 5.3% of Salesforce's customer base per third-party estimates) still trails Salesforce's own headline deal counts
• Exact third-party model providers behind Atlas are not fully publicly disclosed
• Value is largely tied to being on Salesforce's CRM; not usable as a standalone agent platform outside that ecosystem

Security & Privacy

Salesforce documents an "Einstein Trust Layer" that masks personally identifiable information before it reaches underlying models, enforces zero data retention with LLM providers, and logs agent actions for audit; Salesforce also publishes broader Trust and compliance information (e.g., SOC 2, and industry-specific certifications depending on cloud/edition) — consult Salesforce's own Trust and security documentation for the specific certifications applicable to a given org and edition.

Best For

Agentforce fits organizations already standardized on Salesforce CRM (Enterprise Edition or higher) that want to deploy autonomous customer- or employee-facing agents grounded in that CRM data, and that have budget for the additional Data Cloud investment most production deployments require.

Alternatives

• Microsoft Copilot Studio - comparable agent-building platform tied to Microsoft's ecosystem
• ServiceNow Now Assist - agentic AI focused on ITSM/enterprise service workflows
• HubSpot Breeze Agents - comparable agent platform for HubSpot CRM users
• Zendesk AI - customer-service-focused AI agents
• SAP Joule - AI assistant/agents embedded in SAP's enterprise applications

Frequently Asked Questions

Is Agentforce free to use?

A limited entry point exists for free through Salesforce Foundations, which every Enterprise Edition (or higher) customer can access at no extra cost, including Agentforce Builder, Prompt Builder, Agent Script, and a starting allotment of Flex Credits and conversations; broader production use is metered and billed.

How is Agentforce priced?

Salesforce offers several concurrent pricing mechanisms: consumption-based Flex Credits ($500 per 100,000 credits, about $0.10 per action), $2-per-conversation pricing for customer-facing agents, a $125/user/month Agentforce add-on for unmetered employee usage, a $5/user/month Agentforce User License, and bundled Agentforce 1 Editions starting at $550/user/month.

What is the Atlas Reasoning Engine?

Atlas is Agentforce's orchestration layer: it combines multiple large language models, large action models (LAMs), a retrieval-augmented-generation (RAG) module, REST APIs, and data connectors, and is reported to select from roughly 8 to 12 specialized models per query rather than relying on a single model.

Does Agentforce require Salesforce Data Cloud?

Not strictly, but most advanced, production-grade Agentforce deployments rely on Data Cloud (rebranded Data 360) for unified customer data, which independent cost analyses cite as a major driver of total first-year costs beyond the base Agentforce pricing.

Final Summary

Agentforce's technical differentiator is its multi-model Atlas orchestration layer rather than a single flagship LLM, and its pricing is deliberately modular — credits, conversations, and per-user licenses can combine in ways that make total cost highly deployment-specific. Organizations evaluating it should model Data Cloud costs alongside the headline Agentforce numbers, since independent analyses consistently flag that combination as the real driver of enterprise spend.