The most proven Salesforce Agentforce use cases sit in customer service, lead handling, and employee support, and named companies now report results in each. Salesforce packages many of these jobs as seven job-ready AI agents you can turn on and configure.
The Agentforce use cases that work share three traits: lots of volume, steps that repeat, and data the AI agents can actually reach. Picking your first use case is a scoring exercise, not a demo decision, and this guide shows you how to run it.
What Counts as a Salesforce Agentforce Use Case
A Salesforce Agentforce use case is a specific job you hand to AI agents, like answering order-status questions or preparing a wealth advisor for a client meeting. Salesforce’s own library of Agentforce use cases sorts them by function and tags each one by industry.
Think of every use case as four parts: the job, the data it needs, the actions it can take, and the guardrails around it, enforced in part by the Einstein Trust Layer. Miss any one of the four and the use case stalls, usually on data, which is why the connected customer record every agent depends on matters so much.
A use case isn’t the same as an agent. One agent can cover several use cases, and one use case can call on more than one agent, whether it’s prebuilt or built in Agent Builder.
AI agents also aren’t chatbots: a chatbot follows a script and points you to a help article, while an agent reasons over your data and takes action, such as updating a record or starting a process. Salesforce’s own Help site shows the difference, as the service section explains.
Salesforce’s Job-Ready Agents and the Use Cases They Cover
On September 11, 2026, Salesforce added seven job-ready AI agents to Salesforce Agentforce, each built for a named role and each arriving with its own skills, actions, and data model. Several came from the acquisitions behind Salesforce’s newest agents.
| Agent | Job | Status (September 2026) | Use cases it covers | Reported result |
|---|---|---|---|---|
| Casey | Help agent for customer service | Generally available | Service requests across voice, SMS, WhatsApp, and web chat, including returns and escalation | – |
| Paige | IT and HR employee service | Generally available | Internal IT, HR, and admin requests in Slack and employee portals | – |
| Carter | Shopper agent for commerce | Generally available | Product discovery and guided shopping | Hibbett: 90% of core shopper journeys handled within six weeks |
| Hunter | Outbound sales | Pilot, general availability planned for November 2026 | Account research, prospect lists, outreach drafts, multi-week pipeline goals | – |
| Marshall | Supply chain | Generally available | Supply chain operations | – |
| Piper | Inbound pipeline | Generally available | Lead qualification on websites and in inboxes | Asana: 4x conversation volume |
| Fin | Customer experience | Generally available | Complex customer experience workflows across channels | Anthropic: 79% of conversations resolved autonomously |
The Asana result for Piper and the Anthropic result for Fin come from Salesforce’s announcement. The Hunter timeline and the Hibbett result come from Futurum’s analysis of the launch.
Job-ready agents are the “turn it on and configure” option. Anything Salesforce hasn’t packaged you build yourself in Agent Builder, which means designing, building, and testing it before it goes live.
Marshall, the supply chain agent, has the least public detail so far, so treat it as one to watch rather than one to plan around. Casey and Piper are the most practical starting points for most teams, since service and inbound leads are high-volume jobs.
Customer Service Use Cases
Customer service is where Salesforce Agentforce use cases have the longest track record. The Agentforce examples below cover the most common service jobs and what companies report after launch.
Order Status, Returns, and Account Changes
Pandora’s service agent, Clara, handles 45,000 conversations a month, deflects 60% of cases, and lifted Pandora’s net promoter score by 10%. When a request goes beyond what Clara can do, it hands the case to a person with a summary, so the customer doesn’t have to repeat anything.
Clara works because it can see order data from outside Salesforce through Commerce Cloud. That detail matters more than any feature, and the section on what each use case needs covers it.
Troubleshooting and Technical Support
Salesforce runs Agentforce on its own Help site, which gets more than 60 million visits a year and holds more than 740,000 articles. There, Agentforce resolves more than 75% of inquiries in seven languages, replacing earlier Einstein Bots that could list resources but couldn’t personalize an answer.
Energy companies see similar results. According to Salesforce, ENGIE now assists 83% of users with Agentforce on billing and clean-energy questions.
Voice and Contact Center
Casey, one of the job-ready agents built for service, works across voice, SMS, WhatsApp, and web chat. That makes it a natural fit for where contact center automation is heading: fewer handoffs and more requests resolved in one conversation.
Voice costs more to run than text, since each voice action uses more credits. If you’re weighing Casey for a contact center, the cost section below shows how to estimate it.
Sales Use Cases
Sales is the second-most mature area for Salesforce Agentforce. Siemens now qualifies more than 2,500 inbound leads a month for 18,000 sellers using two agents in tandem, and Uber lifted onboarding conversions 3x and processes RFPs 83% faster.
Quoting agents are only as accurate as the rules behind them, so clean pricing rules come before any quoting agent.
Piper handles inbound leads today, and Hunter covers outbound once it’s generally available. For a full walkthrough of sales AI agents, pricing, and rollout order, see our Agentforce for Sales guide.
Marketing and Commerce Use Cases
In commerce, Carter handles guided shopping, and Hibbett’s result above shows how fast a narrow commerce use case can land. Pandora runs a second agent, Gemma, as a digital personal shopper alongside Clara, both built on Commerce Cloud.
On the marketing side, Salesforce reports that 360 MMS cut campaign build time 16x with Agentforce Marketing. Loyalty is a close cousin, and we’ve seen a loyalty program rebuilt for real-time offers once the data flowed properly.
Most commerce use cases depend on live stock and order status. That’s why agents that act on live inventory and order data need integration work before launch, not after.
Field Service Use Cases
Salesforce positions field service AI agents around coordinating schedules, routes, and parts in real time, so technicians stay productive and customers stay informed. These Agentforce use cases fit companies with large technician fleets and frequent rescheduling, and they depend heavily on accurate parts and scheduling data.
Employee, IT, and HR Use Cases
Paige, one of the job-ready agents, handles IT and HR requests in Slack and employee portals, the same kind of internal work that helpdesks spend most of their day on. Requests still respect each employee’s permissions through the Einstein Trust Layer.
Agentforce Coworker is one of three surfaces Salesforce launched with AIforce at Dreamforce 2026. Fulton Bank went from zero to more than 20 Coworker use cases in production in a few weeks, supporting about 3,000 users.
Surfaces like Agentforce Coworker, Slack, and Salesforce in Claude are where people reach AI agents and data, while the agents do the work. Buying Agentforce Coworker doesn’t deploy a service agent, and that split is which is how Claude and Salesforce agents divide knowledge work.
Industry Use Cases
If you’re asking what companies use Salesforce Agentforce, the answer spans banking, telecom, retail, and manufacturing. Here are the Agentforce use cases named companies have published.
Banking and Financial Services
RBC Wealth Management launched a client meeting prep agent in six weeks. It runs from the Household Summary page in Financial Services Cloud, reviews a year of CRM data, and produces a one-page brief for the advisor. Banks building on Financial Services Cloud can reuse that pattern for other advisor and banker AI agents.
Salesforce’s Canadian leadership describes the lesson as starting with narrow, measurable applications rather than sweeping transformation. For banks, transaction-level use cases also need the API layer in place before agents touch payments.
Telecommunications
One NZ built an agent to help hundreds of thousands of prepaid customers move to seven simpler plans. It authenticates each SIM with a one-time password against Communications Cloud, then pulls live balances, add-ons, and usage.
In the US, AT&T uses Agentforce, Data 360, and MuleSoft to give retail experts real-time customer context. Telecom AI agents like these lean on an operational data store that keeps customer context current.
Retail and Consumer Goods
Pandora and Hibbett cover the retail side. Consumer goods companies add trade promotion and retail execution, where agents prioritize store visits and write shelf-reset instructions from compliance and sales data.
Manufacturing and Logistics
Siemens also uses Agentforce to give more than 14,500 partners on-demand guidance, according to Salesforce’s customer stories. In manufacturing, an ERP migration changes what data agents can reach, so time your agent roadmap around it.
In logistics, the groundwork is often a clean sales process first, like a Sales Cloud rollout for a North American logistics provider.
Agentforce Examples: What Named Customers Report
Here are the Salesforce Agentforce examples above in one place, for anyone looking for Agentforce examples with numbers attached. These are company-reported outcomes, sourced in the sections above, and your results will depend on your data and scope.
| Company | Use case | Reported result |
|---|---|---|
| Salesforce (Help site) | Support concierge | More than 75% of inquiries resolved |
| Pandora | Order status and FAQs | 60% deflection, +10% NPS, 45,000 conversations a month |
| ENGIE | Billing and energy questions | 83% of users assisted |
| Siemens | Inbound lead qualification | 2,500+ monthly leads for 18,000 sellers |
| Uber | Lead engagement and RFPs | 3x onboarding conversions, 83% faster RFPs |
| RBC Wealth Management | Advisor meeting prep | One-page brief from a year of CRM data |
| One NZ | Prepaid plan migration | Self-service path for hundreds of thousands of customers |
| Hibbett | Guided shopping | 90% of core shopper journeys |
| Fulton Bank | Employee productivity (Agentforce Coworker) | 20+ use cases, about 3,000 users |
What Each Use Case Needs Behind It
Pandora’s agent is a good reminder of what makes Salesforce Agentforce use cases work. Commerce Cloud pulls fulfilment data from Pandora’s IBM Sterling order management system through direct connectors, and without that connection, “where’s my order” becomes “I don’t know.”
| Use case | Systems outside Salesforce it usually needs | Typical data path | What breaks without it |
|---|---|---|---|
| Order status and returns | Order management, warehouse, carrier | Connectors or MuleSoft APIs | The agent can’t answer the core question |
| Billing and account questions | Billing and payments | MuleSoft APIs | Wrong balances and more escalations |
| Advisor or account prep | Portfolio, core banking, ERP | Data 360 plus MuleSoft | Thin, generic briefs |
| Plan or product changes | Product catalog, provisioning | Industry cloud APIs, such as Communications Cloud | The agent recommends but can’t execute |
| Lead qualification | Web analytics, marketing automation, product usage | Data 360 | Poor prioritization |
| IT and HR requests | HR system, IT service management, identity | APIs or native connectors | Tickets instead of resolutions |
The last two columns reflect how we see these projects play out, not Salesforce documentation. Most gaps trace back to data, and Data 360 is usually the first thing to fix when external data is involved.
Integration for AI agents is different from integration for reports. Agents need live data, filtered by the Einstein Trust Layer and each user’s permissions, at the moment a customer asks, and that’s the difference to test for when you evaluate an integration partner.
What Each Use Case Costs to Run
Salesforce Agentforce actions draw on Flex Credits. Salesforce’s pricing page lists 20 Flex Credits per Agentforce action and 30 per voice action, with credits at $500 per 100,000.
The math is simple: monthly conversations times actions per conversation times 20. For example, 10,000 service conversations with three actions each come to 600,000 credits, or about $3,000 a month at list price.
Your edition includes some credits, and Salesforce lists those allocations per organization per year: 500,000 on Core, 1 million on Advanced, and 2.75 million on Max. The example above would use Core’s full year of credits in under a month.
A new cost line is coming for outside AI agents. Salesforce will count every successful call a registered outside agent makes over the Model Context Protocol (MCP) or the API as a Headless Platform Interaction and charge Flex Credits for it once metering starts.
The rate isn’t published yet, Salesforce promises 30 days’ notice before billing begins, and traditional integrations keep their current pricing. Registered agents also get their own credentials through Agentic Identity, so a Headless Platform Interaction is tied to a specific agent, which makes an audit of which agents can reach which APIs worth doing now.
Before you budget new consumption, check what you already pay for. A quick estimate of your current Salesforce spend often turns up licenses you can redirect.
How to Pick Your First Use Case
Choosing your first use case is where most Salesforce Agentforce programs succeed or stall. Score each candidate from 1 to 5 on four things:
- Volume: how often the job happens.
- Repeatability: how standard the steps are.
- Data readiness: whether the agent can reach what it needs today.
- Cost of a wrong answer: lower risk scores higher.
Here’s an illustration with three common Agentforce use cases.
| Candidate | Volume | Repeatability | Data readiness | Risk (lower is better, scored inversely) | Total |
|---|---|---|---|---|---|
| Order-status questions | 5 | 5 | 4 | 4 | 18 |
| Advisor meeting prep | 3 | 4 | 3 | 4 | 14 |
| Outbound prospecting | 4 | 3 | 2 | 2 | 11 |
Order status wins because it happens constantly, follows the same steps, and fails safely. Outbound prospecting scores lowest because the data is usually scattered and a bad email reaches a real prospect, which is why Piper’s inbound work usually comes first.
The fastest launches above started narrow. RBC and Hibbett each went live in about six weeks because they picked one job, one team, and one measure of success.
Measure two weeks of baseline before launch: containment, first-contact resolution, handle time, and escalation rate. Keep a human in the loop until containment and accuracy hold, and build in guardrails and fallbacks for agents in production from day one.
If you’d rather not score this alone, an assessment does it with your data. Our Salesforce Platform Assessment names your top three to five Agentforce use cases deployable within 90 days, using six readiness dimensions, from data quality to trust and governance.
How Incepta Helps Companies Choose and Deliver Agentforce Use Cases
Incepta works on both sides of what makes Salesforce Agentforce use cases succeed: the agents inside Salesforce and the data outside it. We start with a no-cost AI Operating Model Assessment to find the highest-value use cases and data gaps before any build.
Because Salesforce and MuleSoft work happens in one team, order, billing, ERP, and HR data can reach your agents without a second vendor. Incepta has delivered 1,500+ APIs and 900+ Salesforce implementations, including SAP and ERP integration work.
Our Agentforce QuickStart puts a first production agent live in four weeks. After launch, engineers stay with the agent as data and requirements change, and our AI Catalyst program gives you a path from assessment to production.
Here’s how that maps to the problems above:
- Picking the wrong first use case: a scored shortlist based on value and data readiness.
- Agents that can’t see the data they need: Salesforce and MuleSoft delivered together.
- Consumption and governance surprises: credit modeling, Einstein Trust Layer settings, and agent access audits before launch.
If you’re comparing partners, start with the questions we’d ask of any Agentforce partner, including us. For a quick read on where you stand, take the Agentforce & AI Readiness Scorecard.
The Bottom Line
The best Agentforce use cases start where volume is high, steps repeat, and data is within reach. Launch one narrow first use case with AI agents you can measure, and expand from there.
Frequently Asked Questions
What companies use Agentforce?
Companies using Salesforce Agentforce include Pandora, Siemens, Uber, RBC Wealth Management, One NZ, AT&T, ENGIE, Hibbett, and Fulton Bank. Salesforce also runs Agentforce on its own Help site, where it resolves more than three-quarters of inquiries. These Agentforce examples range from order status and lead qualification to advisor meeting prep and prepaid plan changes.
What can Agentforce do for my business?
Salesforce Agentforce AI agents can answer and resolve customer service requests, qualify inbound leads, prepare staff for customer meetings, guide online shoppers, handle IT and HR requests through Paige or Agentforce Coworker, and support industry jobs like plan changes or claims. The right starting point depends on your volume and data, so scope the work before you hire help.
What is the best first Agentforce use case?
The best first Agentforce use case has high volume, repeatable steps, data the agent can already reach, and a low cost when it gets something wrong. For most companies, that’s order status or account questions handled by Casey, or internal IT requests handled by Paige. Outbound prospecting is usually a poor first choice because the data is scattered. Start with a human in the loop and widen autonomy as accuracy holds.
Does every Agentforce use case need Data 360?
No. Agentforce use cases that only use Salesforce data, like opportunity updates or internal knowledge answers, run without Data 360. You need Data 360 when AI agents rely on external or unstructured data such as web behavior, product usage, or records from other systems. Many customer-facing use cases reach that point quickly.
How much does an Agentforce use case cost to run?
A Salesforce Agentforce use case costs 20 Flex Credits per standard action and 30 per voice action, with credits listed at $500 per 100,000. Multiply conversations by actions per conversation by 20 to estimate credits. Salesforce also plans to meter calls from registered outside agents as a Headless Platform Interaction, at a rate not yet published.
How long does it take to launch a first Agentforce use case?
A narrow first Agentforce use case can launch in weeks. RBC Wealth Management and Hibbett each went live in about six weeks, and Incepta’s QuickStart puts a first production agent live in four. Broader rollouts across teams and systems usually take several months, mostly because of data and integration work.
Is Agentforce better for B2C or B2B companies?
Salesforce Agentforce has the longest record in high-volume B2C service, like Pandora’s order-status agent. B2B works well too when the data is connected, as Siemens shows with lead qualification and RBC with advisor prep. The deciding factor is less the business model and more whether the agent can reach the right data.
Can Agentforce agents act on systems outside Salesforce?
Yes. Agentforce agents can read from and act on outside systems through connectors, MuleSoft APIs, and the Model Context Protocol, as Pandora’s agent does with its separate order management system. Registered outside agents calling Salesforce will soon consume Flex Credits for each successful call, and each needs its own Agentic Identity. Access still runs through the Einstein Trust Layer and your permissions.
