Salesforce Agentforce Review (2026)
Salesforce's AI agent layer, which makes sense in proportion to how much of your business already runs on Salesforce.
Some links on this page can be affiliate links. We may earn a commission at no cost to you. This never affects our rankings. How we work.
Visit Salesforce Agentforce ↗Opens www.salesforce.com in a new tab.
What the score measures. Five things we can verify from published material: whether pricing is transparent, whether the product can act on your systems or only answer from documents, whether the vendor still owns its own roadmap, how much independent review evidence exists, and whether the bill stays predictable as volume grows. Weighted, then scored against an ideal platform that scores 5.
BEST FOR: SALESFORCE-STANDARDISED ORGANISATIONS · LAST UPDATED 2026-07-30
Salesforce Agentforce in depth
Agentforce is Salesforce's answer to autonomous AI agents, and its argument is data proximity: the agent runs where your customer records, business logic and permissions already live, so it does not need integration to know who it is talking to.
For an organisation genuinely standardised on Salesforce, that is a real advantage. The hardest part of deploying support AI is usually giving it safe access to your systems, and if your systems are Salesforce, much of that work is already done.
What it does
- AI agents built on Salesforce data and metadata
- Native access to CRM records and business logic
- Agent Builder for configuring autonomous actions
- Consumption-based pricing on actions
- Service Cloud integration
- Data Cloud grounding
More detail on how it works
Salesforce prices Agentforce on consumption — per action rather than per seat — which is a third billing model to add to the per-seat and per-resolution options elsewhere in this category, with its own forecasting difficulties.
Data proximity
The genuine advantage. The agent reads CRM records, respects your existing permission model and uses business logic you have already built. No middleware, no separate identity layer.
Agent Builder
Configures what the agent may do, using Salesforce's declarative tooling. Familiar to your admins, which matters — the people who will maintain it already know the environment.
Consumption pricing
Billed on actions. Like per-resolution billing, cost rises with use, so model it against your volume carefully. Salesforce consumption pricing has a reputation for being hard to forecast and that reputation is earned.
Setting Salesforce Agentforce up
Realistic time to a working deployment: Months. These are the standard steps for this category. We have not published a walkthrough specific to this tool yet.
- 01
Scoping and statement of work
Weeks with the vendor or a partner before anything is built. Implementation is usually a separate line item from licences.
- 02
Telephony and channel migration
Numbers, carriers, routing, recording and compliance retention. Infrastructure work with lead times.
- 03
Integrate CRM and back office
The bulk of the engineering. Agent desktop, screen pop, and whatever the AI needs to read.
- 04
Configure the AI layer
Increasingly bundled but usually a separate licence. Confirm which AI features your contract actually includes.
- 05
Train agents and supervisors
New desktop, new workflows, new reporting. Budget real hours, not a webinar.
- 06
Phased cutover
By queue or by region. Nobody moves a whole contact centre in one weekend twice.
The step that takes longer than they imply. Everything. Contact centre platforms are implementation projects with statements of work, not software you switch on.
What the AI actually does
Every platform here says “AI agent”. It covers a bot that reads your help centre and a system that can refund a customer. This is which one you are buying.
Agentforce acts only through things that already exist in your org, and every one of them has to be exposed on purpose.
An action is an Apex class, a SOQL query, a flow or a prompt template that somebody deliberately published to the agent.
That makes its reach unusually knowable, and unusually dependent on Salesforce development capacity you may not have.
- Four documented ways to build one: Apex REST classes turned into an OpenAPI document, Apex controller methods marked AuraEnabled, Named Query APIs exposing custom SOQL, and Apex methods annotated with @InvocableMethod
- Nothing is available by accident. A Named Query API has to be manually activated in the org's API Catalog before an agent can use it at all
- Permissions are separate from capability: creating Named Query API records needs create, read, update and delete rights, and using an agent built on them needs View Developer Name or View Setup and Configuration
- Custom actions can be Apex, flows or prompt templates, so the same mechanism covers reading a record, writing one and generating text, and the difference is in what you exposed rather than in the agent
- The credit price is published even though the per-conversation rate is not: Flex Credits at $500 per 100,000, which is half a cent each, against Service Cloud seats at $25, $100 and $175
Who Should Use It
The second list is the more useful one.
Salesforce-standardised organisations
Where CRM, service and business logic already live in Salesforce, the data proximity advantage is genuine and large.
Teams with Salesforce admin capacity
Agent Builder uses tooling your admins know, which makes ongoing maintenance realistic.
Enterprises wanting one vendor
Consolidating the agent with the CRM removes an integration and a supplier relationship.
Who should look elsewhere. Anyone not on Salesforce. Without the data proximity the advantage disappears entirely, and specialist agents are cheaper, faster to deploy and better at resolution. Also difficult for teams that need a predictable bill, given consumption pricing.
Strengths and Weaknesses
What Works
- Genuine data proximity advantage inside Salesforce.
- Uses existing permissions and business logic.
- Declarative tooling your admins already know.
What Does Not
- Only compelling if you are standardised on Salesforce.
- An action is not a resolution, which makes comparison against competitors awkward.
Salesforce Agentforce pricing
Consumption-based, priced per action, on top of your existing Salesforce licensing. Rates are quoted rather than simply published in a form you can act on.
The forecasting problem is real: if you cannot predict your action volume, you cannot predict your bill. Ask for a modelled estimate against your own ticket data and get the per-action definition in writing.
The full arithmetic
The comparison against Fin at $0.99 per resolution is not straightforward, because an action is not a resolution — one resolution may consume several actions. Establish the conversion before comparing.
Custom and private pricing
Salesforce Agentforce does not publish a rate. You have to ask, which means a sales conversation before you can compare it against anything else on this site.
Contact sales ↗Ask two things on that call: what triggers a billable event, and what the rate is at twice your current volume.
INTEGRATIONS
Our Recommendation
Who we would tell to buy this, and who we would not.
Agentforce is a strong option in exact proportion to how much of your business runs on Salesforce. If your customer records, permissions and business logic already live there, the agent starts with safe access to all of it, and that is normally the hardest and most expensive part of deploying support AI. Two cautions. Consumption pricing on actions is genuinely hard to forecast — get a model against your own data and the per-action definition in writing. And if you are not a Salesforce shop, none of the advantage applies and specialist agents will beat it on cost and resolution.
Where we write about Salesforce Agentforce
2 pieces on this site name it.
- One vendor prints how long a deployment really takes. It is sixty days, for the simplest product here.We counted the documented setup steps at seven platforms and put them next to the time each vendor claims. Two publish no documentation at all. And at three of them, the testing you do before launch is billable.
- The vendor ships the reasoning. You ship the reach.Fourteen platforms, read from their own documentation, split by who has to build the thing the agent acts through. It decides your implementation cost.
Similar Tools
Others in enterprise & ccaas.
Frequently Asked
The questions people actually search for about Salesforce Agentforce.
What is Salesforce Agentforce?
Salesforce's AI agent layer, running on your CRM data with native access to records, permissions and business logic already in Salesforce.
How much does Agentforce cost?
Consumption-based, priced per action, on top of existing Salesforce licensing. Ask for a model against your own volume and get the per-action definition in writing.
Is Agentforce better than Intercom Fin?
Inside a Salesforce-standardised organisation, the data proximity is a real advantage. Outside one, Fin is cheaper, faster to deploy and generally better at resolution.
Do I need Salesforce to use Agentforce?
Effectively yes. The entire advantage is that the agent runs where your data already lives.
How does consumption pricing work?
You are billed per action the agent takes. One resolution may consume several actions, which makes comparison against per-resolution vendors awkward until you establish the conversion.
Can Agentforce take actions in my systems?
Yes, using Salesforce Flows, Apex and your existing business logic — with your existing permission model applied.
What are the main Agentforce alternatives?
Fin, Ada or Decagon as specialist agents, Zendesk AI if you run Zendesk, or Microsoft's equivalent if you are standardised on Dynamics.