Revenue Cloud quoting agent: what it can do today

The quoting agent in Revenue Cloud turns a request in plain language into a quote: it picks valid products from the catalog, prices them with Salesforce Pricing, applies discounts within the guardrails and can amend and summarise quotes. For an internal sales desk with standard products and clean pricing data, that saves real time. Before it goes live, you need a maintained catalog, synced prices, configuration rules in Salesforce and discount limits that approvals enforce. For configured machines whose rules live outside Salesforce, the agent cannot help yet.

What does the Revenue Cloud quoting agent do?

Salesforce introduced agent quoting for Revenue Cloud in July 2025. In the announcement1, reps describe what they need, and the agent generates the quote, “automatically pulling the correct products, pricing, and terms”. The headline numbers come from Salesforce’s own sales team: 75 percent less quoting time and 87 percent fewer clicks. That is a vendor figure from one company and its own sales team. I would not put it into a business case for a machine builder.

The Trailhead quick start for a quoting agent2 is more precise about the scope. The agent:

• creates, modifies and summarises quotes and respects role-based access,
• follows product catalog and fulfillment rules, so it quotes only valid products,
• keeps pricing and discounting within system guardrails,
• handles new business quotes, amendments to assets and quote updates.

The work is done by the Quote Management subagent, whose actions are based on flows. In the exercise you ask for 15 headsets with 15 percent discount for an account and opportunity, then change the discount to 10 percent; you then email the quote yourself from the opportunity with a template. Salesforce Help lists a second building block, the Product Selection subagent, for searching and filtering products. The product is called Revenue Cloud on salesforce.com; Help and Trailhead file the agent under Agentforce Revenue Management or Revenue Management.

Catalog, prices and rules come before the agent

The agent reads the same records your sales desk reads. What an experienced rep corrects by habit, the agent passes on. So the preparation is mostly data and rules work:

• Catalog. Every sellable product, option and spare part as a catalog entry with an active flag and a description a rep would type. How to model variants without copying the ERP material master is covered in the chapter catalog, configurator, pricing.
• Synced pricing data. The Trailhead setup starts with Sync Pricing Data in Salesforce Pricing Setup. If prices are not synced, the agent quotes nothing useful.
• Configuration rules in Salesforce. The agent stays within catalog rules. If your rules sit in SAP variant configuration or in an engineer’s head, the agent does not see them.
• Discount limits as approvals. The agent may apply a discount a rep asks for. Whether it needs approval should be decided by the approval workflow, as described in discount approvals in Revenue Cloud.
• Agentforce setup. Agentforce enabled, an employee agent with the Quote Management subagent and a permission set that grants agent access.

I would count the share of products that a rep can find by the words customers use. If it is low, fix that before the pilot.

Winter ’27 adds self-service, partners and renewals

The Winter ’27 announcement3, generally available since 12 October 2026, extends agent quoting to three groups: “self-service quoting for customer buyers, integrated quoting for partners, and automated renewal package assembly for account executives”. Customers “order without a rep”, partners “transact without leaving their workspace”, and account executives get “a drafted quote and risk signals” before a renewal.

For manufacturers, renewal drafts are the most interesting part. Maintenance and remote monitoring subscriptions renew every year, and preparing the renewal quote is routine work. Partner and self-service quoting raise commercial questions first: which prices a dealer sees, and whether a customer’s self-service quote may undercut the dealer.

If your dealers quote in Manufacturing Cloud instead, look at the AI Partner Sales Assistant for dealer quotes. Two quoting agents on two quote models is the setup I would avoid.

What does the quoting agent cost?

Two cost blocks apply. First the Revenue Cloud licences for the users the agent works for. Second the agent’s consumption in Flex Credits, at the rates on the Agentforce pricing page4:

$500

per 100,000 Flex Credits

20

Flex Credits per standard action

$0.10

per standard action

Prompts come on top and depend on the model tier.

A rough estimate, as an assumption: a quote dialogue needs five actions, such as finding the account, searching products, creating the quote, adding lines and applying a discount. That is 100 credits or $0.50 per quote. At 1,500 agent quotes a month you are at about $750 plus prompts. Tests in a sandbox also consume credits. How to estimate all meters is in Flex Credits: estimating Agentforce cost per action, and the AI masterclass chapter on costs and tokens puts it into a budget.

When is the quoting agent the wrong tool?

I would not use it, or not yet, in four situations:

1. Engineered to order. If each machine is designed for the order, the quote comes from engineering, not from a catalog.
2. Rules outside Salesforce. If configuration logic sits in SAP variant configuration or a PLM, the agent quotes combinations that cannot be built.
3. Prices only in SAP conditions. If Salesforce holds list prices and the net price comes from SAP, the agent quotes the wrong price unless pricing calls SAP.
4. Few quotes with high value. Ten quotes a month with weeks of negotiation do not need an agent; they need a good sales engineer.

Worked example: spare parts and retrofit quotes on the sales desk

A maker of packaging machines runs a sales desk that answers spare parts and retrofit requests from existing customers: wear part kits, upgrade packages and service hours. Prices are in Salesforce price books, synced from SAP every night, and the catalog has clean descriptions. The desk handles a large number of small quotes.

Here the agent fits well. A rep types “wear part kit for line 3 plus two service days, usual discount”, and the agent finds the kit and the service item in the catalog and builds the quote for the customer’s account. Discounts above the limit go to the approval workflow. New machine quotes stay with the sales engineers, because their configuration rules still live in SAP. The agent is introduced in one country first and tested with real requests from the last quarter.

Write 30 test requests for one quote type

Pick one quote type with standard products and synced prices, write 30 test requests from real cases and define the expected quote for each. Run them in a sandbox before any customer or dealer sees an agent quote, as described in testing Agentforce agents before go-live. Where agents fit in manufacturing beyond quoting is covered in the chapter AI agents in manufacturing, and the whole path from quote to order in quote, contract, order.

Sources4

1. 1announcement (opens in a new tab)salesforce.com
2. 2Trailhead quick start for a quoting agent (opens in a new tab)trailhead.salesforce.com
3. 3Winter ’27 announcement (opens in a new tab)salesforce.com
4. 4Agentforce pricing page (opens in a new tab)salesforce.com