Best AI Agents on Salesforce AgentExchange 2026: Top Picks by Role

Finding the best AI agents Salesforce AgentExchange 2026 has to offer takes more effort than you'd expect. The marketplace has matured quickly since its 2025 launch, and by early 2026 there are enough genuinely useful prebuilt agents to make curation worthwhile — but also enough noise to waste your evaluation time. Each entry covers what the agent actually does, what license it requires, and what setup honestly looks like. The parts most guides skip are the parts we lead with.

If you're an admin or ops leader evaluating Agentforce deployments, use this list to prioritize. Most agents here are reactive: they respond when triggered. For the newer proactive, event-driven architecture, see our deep-dive on how Agentforce 2dx reshapes the AI platform.

---

What Is AgentExchange?

AgentExchange is Salesforce's dedicated marketplace for prebuilt Agentforce agents, agent templates, actions, and topics. Think of it as AppExchange built for autonomous AI — but keep in mind you'll find both complete prebuilt agents and reusable components (individual actions and topic packages) that you can add to agents you're already building. Not every listing is a turn-key autonomous agent; some are building blocks, and knowing the difference before you click "Get It Now" saves confusion during setup.

You can reach it two ways:

• Directly: agentexchange.salesforce.com
• From Setup: Type Agents in the Quick Find box, then select Agentforce Agents; an AgentExchange tab is available from within that interface.

Agents listed there are built using the same primitives you'd use yourself — topics, actions, and instructions — but the heavy lifting of prompt engineering, action wiring, and testing has already been done by Salesforce or an ISV partner. You browse, click Get It Now, connect to your org, and then finish customization in Agentforce Builder (the unified agent configuration UI in Setup — previously called "Agent Builder," renamed as of Spring '26).

Some agents are free with your existing Agentforce licenses. Others are ISV-published and carry separate pricing. We'll flag which is which for each entry below.

---

Best AI Agents on Salesforce AgentExchange 2026, by Role

Sales

---

1. SDR Agent (Salesforce)

Publisher: Salesforce
Role: Sales Development Representative support
Core Use Case: Handles inbound lead qualification autonomously — responds to web form submissions and inbound email leads, asks qualification questions over email or chat, scores responses against your ICP criteria, and books meetings directly onto rep calendars via Salesforce Scheduler. Reps wake up to qualified meetings rather than a queue of cold leads to chase.
Pricing: Included with Agentforce for Sales license (no additional charge beyond base license cost)
Setup Complexity: Moderate — requires connecting Salesforce Scheduler, configuring your ICP qualification criteria as agent instructions, and mapping lead fields; plan on two to four hours for a clean org.

---

2. Sales Coach Agent (Salesforce)

Publisher: Salesforce
Role: Account Executive / Sales Rep coaching
Core Use Case: Provides deal-based coaching and AI-driven pitch role-play simulations for sales reps. Sales Coach analyzes open opportunities and coaches reps on how to approach a specific deal — surfacing talking points, flagging risks, and generating practice scenarios against simulated buyer personas. Note: talk-to-listen ratio analysis, competitive mention tracking, and transcript-level metrics are features of Einstein Conversation Insights, a separate product. If you want both deal coaching and call analytics, both products are needed and both require separate licensing. Managers configure coaching rubrics; reps get guided prep before calls, not just feedback after them.
Pricing: Part of the Agentforce for Sales license bundle; however, requires separate Einstein Agent user license provisioning per user (Setup → Company Information → Licenses). "Included with Agentforce for Sales" does not mean licenses are automatically assigned — confirm seat availability with your AE before rollout, as orgs frequently discover they have fewer agent seats than expected.
Setup Complexity: Moderate — core Sales Coach setup is manageable if your opportunity data is clean and stage usage is consistent. If you also want Einstein Conversation Insights call analytics, telephony integration is a separate project of its own.

---

3. Opportunity Qualification Advisor (Salesforce Partner ISV — varies by listing)

Publisher: Listed ISV partner (multiple vendors offer variants; check AgentExchange search for "opportunity qualification")
Role: Account Executive / Sales Manager
Core Use Case: Analyzes open pipeline opportunities against historical win/loss patterns stored in your org, flags deals that show signs of stall (no activity in N days, missing next step, single-threaded contact engagement), and recommends specific next actions. Unlike a dashboard, this agent proactively surfaces the risk in a Slack message or Chatter post without a manager having to pull a report.
Pricing: Pricing not publicly listed — contact vendor; most ISV agents in this category are licensed per org or per user/month
Setup Complexity: Complex — requires clean historical opportunity data for pattern analysis; orgs with inconsistent Stage usage or sparse activity logging will get poor recommendations until data hygiene is addressed.

---

Service

---

4. Service Agent (Salesforce)

Publisher: Salesforce
Role: Customer Service Representative
Core Use Case: Handles Tier-1 case deflection across chat and email channels — answers questions by pulling from your Knowledge articles, checks order status via connected flows, processes routine requests like password resets or return initiations, and escalates to a human rep with full context when it can't resolve. The handoff quality (passing the conversation history and a case summary) is notably better than prior bot frameworks.
Pricing: Included with Agentforce for Service license
Setup Complexity: Moderate — Knowledge article quality is the key variable. Orgs with well-structured, up-to-date Knowledge bases get good deflection rates quickly; orgs with stale or sparse articles will need a Knowledge audit first.

Evaluating a broader AI-native contact center deployment rather than targeted case deflection? Our Agentforce Contact Center guide covers the full telephony-integrated, omni-channel setup and whether it should replace your existing CTI platform.

---

5. Field Service Scheduling Agent (Salesforce)

Publisher: Salesforce
Role: Field Service Dispatcher / Customer
Core Use Case: Handles inbound appointment requests for field service visits — customers can request, reschedule, or cancel appointments via a chat or self-service portal. The agent checks technician availability via Field Service (FSL) optimization rules, proposes time windows, confirms the booking, and sends reminders. Dispatchers see the confirmed appointments without having to manually work inbound scheduling requests.
Pricing: Included with Agentforce for Service + Field Service license; FSL required
Setup Complexity: Complex — FSL territory management, service territories, and scheduling policies must already be configured correctly; this agent is a front-end layer on top of FSL, not a replacement for FSL setup.

---

6. IT Help Desk Agent (Salesforce / ISV)

Publisher: Salesforce (base template); ISV variants available
Role: Internal IT Support / Employee Self-Service
Core Use Case: Handles common employee IT requests — software access provisioning, password resets, VPN troubleshooting, and hardware request routing — via an internal Slack or Experience Cloud deployment. Integrates with your IT ticketing flow in Service Cloud, auto-creates cases for issues it can't resolve, and assigns to the right IT queue based on issue category.
Pricing: Base template included with Agentforce Platform license; ISV variants (e.g., with deeper ITSM integrations) priced separately — contact vendor
Setup Complexity: Moderate — the base template is quick to deploy if you're already using Service Cloud for IT cases; custom integrations to Jira, ServiceNow, or LDAP add complexity.

---

HR

---

7. HR Employee Services Agent (Salesforce)

Publisher: Salesforce
Role: HR Business Partner / Employee
Core Use Case: Answers employee questions about HR policies, benefits, PTO balances, and onboarding checklists by pulling from a curated set of HR Knowledge articles and connected HR system data. Deployed on an internal Experience Cloud portal or Slack, it reduces HR ticket volume for routine policy questions. For requests it can't answer (e.g., leave of absence requests requiring human judgment), it routes to the appropriate HR case queue with context.
Pricing: Included with Agentforce Platform license; HR Cloud or a connected HRIS integration may be required for live data (PTO balances, benefits enrollment status)
Setup Complexity: Moderate — requires building out an HR Knowledge base in Salesforce (or connecting an existing one), and configuring data flows from your HRIS for live data queries.

---

8. Recruiting Coordinator Agent (ISV Partner)

Publisher: ISV Partner (search "recruiting" or "talent acquisition" on AgentExchange)
Role: Talent Acquisition / Recruiting Coordinator
Core Use Case: Handles candidate scheduling logistics — coordinates interview slots between candidates and hiring managers by checking calendar availability, sends interview confirmation emails, collects pre-interview assessments, and follows up with candidates who haven't responded. Recruiting coordinators spend a disproportionate amount of time on scheduling logistics; this agent handles the mechanical parts so they can focus on candidate experience.
Pricing: Pricing not publicly listed — contact vendor; typical ISV licensing in this space is per seat or per requisition volume
Setup Complexity: Moderate — calendar integration (Google Workspace or Microsoft 365) and connection to your ATS are required; if your ATS is already integrated with Salesforce Recruiting, setup is faster.

---

Finance & Ops

---

9. Collections & Billing Inquiry Agent (ISV Partner)

Publisher: ISV Partner (search "collections" or "billing" on AgentExchange)
Role: Accounts Receivable / Finance Operations
Core Use Case: Handles inbound billing inquiries from customers — invoice disputes, payment confirmation requests, and payment plan requests — via chat or email. Pulls invoice data from Revenue Cloud or a connected ERP, confirms payment status, and can initiate payment plan workflows. For disputes requiring human review, it creates a case with the invoice details already attached, cutting AR team handling time.
Pricing: Pricing not publicly listed — contact vendor
Setup Complexity: Complex — requires clean Revenue Cloud or ERP integration; invoice data quality and payment system connectivity are prerequisites, and finance teams typically require rigorous UAT before go-live.

---

10. Procurement & Vendor Inquiry Agent (ISV Partner)

Publisher: ISV Partner (search "procurement" or "vendor management" on AgentExchange)
Role: Procurement / Vendor Management
Core Use Case: Handles vendor-facing inquiries about PO status, payment timing, and contract renewal reminders. Vendors can query a self-service portal powered by this agent to check where their purchase order stands or when their payment is expected to process, without calling your AP team. Internally, procurement staff can ask the agent to summarize a vendor's contract history or flag contracts expiring within 90 days.
Pricing: Pricing not publicly listed — contact vendor
Setup Complexity: Moderate to Complex — depends on whether your PO and contract data lives in Salesforce natively or requires integration with a procurement system (Coupa, Ariba, etc.).

---

How to Install an AI Agent from Salesforce AgentExchange

This walkthrough uses the Service Agent as the example — it's the most commonly deployed and the clearest path for first-timers.

Step 1: Navigate to AgentExchange

From within your Salesforce org, go to Setup and type Agents in the Quick Find box, then select Agentforce Agents. From the Agentforce Agents page, click the AgentExchange tab to browse and install from within your org context. Alternatively, navigate directly to agentexchange.salesforce.com in a browser and log in with your Salesforce credentials — both paths lead to the same listings.

Step 2: Find the Agent

Use the search bar or browse by category. For Service Agent, filter by Category: Service or simply search "Service Agent." You'll see the listing with a description, screenshots, and any listed requirements.

Review the Requirements section carefully before proceeding — it will list required licenses, connected products (e.g., Data Cloud, Knowledge), and any Salesforce edition requirements.

Step 3: Click "Get It Now"

The experience varies by listing type. Full ISV agent packages follow the standard AppExchange package install flow — click Get It Now, choose Install in Production or Install in Sandbox (always use sandbox first), and log in to the target org when prompted. Accept the permissions requested. However, Salesforce-native actions, topics, and agent templates may install directly into Agentforce Builder without the package wizard — you'll be taken straight into your org's agent configuration UI instead. Either way, always start with sandbox.

Spring '26 note: Agentforce Builder was renamed from "Agent Builder" in the Spring '26 release. If your org is on an older release, your navigation labels will differ — see the Salesforce Spring '26 complete admin guide for a full breakdown of the UI and naming changes.

Step 4: Post-Install Configuration in Agentforce Builder

After installation, you'll be redirected (or can navigate) to Setup → Quick Find: Agents → Agentforce Agents. You'll see the newly installed Service Agent listed. Click Open in Agentforce Builder (or Edit depending on your release version).

Inside Agentforce Builder you'll see:

• Topics — the agent's defined areas of responsibility (e.g., "Handle Case Inquiry," "Process Return Request"). Review these and disable any topics that don't apply to your org.
• Actions — the specific functions the agent can execute (e.g., "Search Knowledge," "Create Case," "Check Order Status"). Each action maps to a Flow, Apex class, or API call. You'll need to ensure the underlying Flows are activated and any referenced objects are accessible.
• Instructions — natural language rules that govern the agent's behavior. Edit these to add org-specific constraints (e.g., "Do not discuss competitor products" or "Always confirm the customer's email before creating a case").

Step 5: Configure the Channel

Go to Channels within the agent configuration to specify where the agent is deployed — Messaging (chat), Email, or a custom channel. For Service Agent, this typically means connecting to an existing Messaging channel (configured under Setup → Messaging Settings).

Step 6: Test in Preview

Use the Preview pane inside Agentforce Builder to run test conversations before activating. Test your most common service scenarios and at least one escalation path to verify handoff behavior. Check that cases are being created with the correct record type, owner, and queue assignment.

Step 7: Activate and Monitor

When testing looks solid, toggle the agent to Active and monitor early conversations using the Agentforce monitoring tools in Setup. Look for an Events tab or observability dashboard within the Agentforce Builder — the exact UI labels for these monitoring features are evolving across releases, so check the current Setup navigation for "Agentforce Observability" or "Agent Activity" in your org's version. Set up an escalation alert if case deflection drops below your target threshold.

---

Licensing and Prerequisites at a Glance

Before you install anything, confirm you have the right license in place. The table below summarizes what each agent category requires. License SKU names shown in Setup and on your Salesforce contract may vary — always check Setup → Company Information → Licenses for the exact names provisioned in your org, as Spring '26 rebranding has shifted some product labels (see the Salesforce Spring '26 complete admin guide for the full picture).

Edition gate: Agentforce is only available on Enterprise, Performance, Unlimited, and Developer editions. Professional edition orgs cannot use Agentforce — this is a hard prerequisite that the listing requirements may not surface prominently.

Free evaluation tier: Orgs without Agentforce licenses can use Agentforce Foundations (free, limited conversation credits) for evaluation and proof-of-concept work before committing to a paid license. Ask your AE to enable it.

---

Org Prerequisites: What You Need Before Installing

Not every Agentforce agent will work in every org. Here's what to confirm before you commit time to an installation:

Einstein Must Be Enabled First

Before any agent install, confirm Einstein is enabled in your org (Setup → Einstein Setup) and that your org is running a recent API version. Many orgs — especially those on older Enterprise edition setups that never activated Einstein features — hit this wall post-install when the agent won't initialize. This takes five minutes to check and can save hours of debugging.

Agentforce Licenses

Most agents require one of the following:

• Agentforce for Sales — for sales-focused agents (SDR Agent, Sales Coach)
• Agentforce for Service — for service and field service agents
• Agentforce Platform — a more general license that covers IT help desk, HR, and internal-facing use cases

Check your org's Company Information page (Setup → Company Information) and confirm with your AE if you're unsure which licenses are provisioned. Running out of license seats mid-rollout is a common and avoidable problem.

Agent User Record

Create a dedicated Salesforce user record specifically for the agent — this is the identity the agent uses to interact with your org, not a person's login. Assign the appropriate Einstein Agent license to this user record and grant the Agent User permission set. This step is documented but easy to overlook, and missing it is one of the most common reasons a newly installed agent silently fails to respond.

Data Cloud

Several agents — particularly those doing personalization, next-best-action, or cross-channel context assembly — require Data Cloud (formerly Customer Data Platform). The Service Agent in its base form does not require Data Cloud, but advanced configurations (e.g., surfacing unified customer profile data during a chat) do. Read the requirements section of each AgentExchange listing carefully.

Sandbox Requirements

Always install in a Full Copy or Partial Copy sandbox before production. Developer sandboxes can be used for initial exploration but often lack the data volume needed to validate agent behavior realistically. Pay particular attention to:

• Knowledge articles being present and published in the sandbox
• Flows referenced by agent actions being activated
• Permission sets granted to the agent user (agents run as a named user with a specific permission set — this needs to be configured in the sandbox exactly as it will be in prod)

API and Integration Credentials

If an agent calls external systems (ERP, HRIS, telephony), those API credentials need to be configured as Named Credentials in each org environment. Don't forget to update Named Credentials when moving from sandbox to production.

---

What Nobody Tells You Before You Install

The AgentExchange listing page will tell you what an agent does. It won't tell you where it falls apart in your specific org. Here are the friction points practitioners actually hit:

1. Agent Permissions Don't Configure Themselves After Install

After a package installs successfully, it's tempting to go straight to Preview and start testing. Nine times out of ten, the agent won't respond — not because of a config issue, but because the agent user has no permission to access the objects your actions are trying to touch. After install, you need to explicitly verify that the agent's dedicated user has: the Einstein Agent license assigned, the Agent User permission set granted, read/write access on every object the agent's actions reference, and access to the correct Knowledge data categories. This is a checklist, not a one-click operation.

2. Agents Don't Travel via Change Sets

When you're satisfied with your sandbox configuration and ready to promote to production, do not reach for Change Sets. Agent metadata — agent definitions, topics, actions, instructions — does not live in the standard Change Set component library. You need a metadata API deployment using Salesforce CLI (sf project deploy start) or a second install directly from AgentExchange into your production org (followed by manually replicating your configuration). Admins who discover this at 10pm before a go-live have not had a good night. Plan the deployment path before you start configuring.

3. "Prebuilt" Means the Skeleton, Not the Finished Agent

What AgentExchange typically delivers is a template: the topic structure, sample action references, and a starting set of instructions. It is not a production-ready agent. Before it will work in your org, you still need to activate the underlying Flows that actions call, map the correct object API names to action inputs, write the final natural-language instructions for your org's specific policies and constraints, connect it to a channel, and run end-to-end tests with realistic data. Calling these "prebuilt" is accurate in the sense that the architecture is done. The org-specific configuration is yours.

4. Your Data Problems Become the Agent's Problems — Immediately

Agents expose data quality gaps that humans quietly worked around for years. Common examples: Contact records with no email address cause the SDR Agent to fail mid-qualification sequence; Opportunity records with empty Next Step fields give Sales Coach nothing to evaluate; Knowledge articles with outdated product names or broken image links produce confusing Service Agent responses; Account records missing a billing region break scheduling logic in the FSL Scheduling Agent. The agent has no ability to shrug and move on the way a human rep would. Run a targeted data quality audit against the specific fields each agent's actions will read before you go live — not after users start complaining.

5. Einstein Trust Layer Has Data Residency Implications You Should Confirm Early

All Agentforce agents route prompts and context data through Salesforce's Einstein Trust Layer, which governs how data is sent to the underlying LLM (including zero-data-retention guarantees with third-party model providers). For orgs in regulated industries — healthcare, financial services, public sector — confirm with your Salesforce AE which data residency regions are available for your specific edition and Hyperforce configuration. Not all Hyperforce regions support all Agentforce features at the same release cadence, and discovering a regional gap three weeks before a planned launch is painful. Get this answered in the discovery call, not the go-live review.

6. Conversation Credits Are a Finite Resource

Each agent interaction consumes Agentforce conversation credits. Billing is typically per conversation session (not per message), but high-volume deployments can burn through a credit allocation faster than expected. If you're deploying a Service Agent on a busy support channel and your license includes a fixed credit pool, model your expected monthly conversation volume before launch and confirm the credit allocation is sufficient. A mid-month hard stop on agent conversations — with no human fallback configured — is a support incident.

---

Closing Thoughts

AgentExchange in early 2026 is past the "shiny demo" stage. The Salesforce-native agents — particularly Service Agent, SDR Agent, and Sales Coach — are production-ready for orgs that have clean data and the right licenses. The ISV catalog is growing but more variable in quality; note that Salesforce does conduct a security review (similar to AppExchange) for listed packages, but functional quality and implementation depth are still buyer-beware. Check Trailblazer Community reviews on each listing before committing to an ISV agent. Always run a sandbox POC with realistic data before going to production.

The biggest mistake we see orgs make is treating AgentExchange like a one-click solution. The agents handle the AI reasoning layer well. The bottlenecks are almost always upstream: stale Knowledge articles, dirty pipeline data, half-configured Flows, missing integrations, or overlooked permission assignments. Fix those first, and the agents perform. Skip that work, and no prebuilt agent will save you.

Next steps:

• Audit your org for the prerequisites listed above before browsing AgentExchange
• Confirm your edition (Enterprise or above) and Agentforce license type before evaluating specific agents — use the licensing table above as your starting guide
• Start with one agent in a single department — Service Agent for a support team or SDR Agent for an inbound BDR function are the lowest-risk starting points
• After deployment, monitor performance in Agentforce Observability / Command Center, not a generic reporting dashboard
• Join the Trailblazer Community's Agentforce group to see what orgs similar to yours are deploying and where they hit friction

Have you deployed an agent from AgentExchange? We'd like to hear what worked and what didn't — reach out via the CRM×AI contact page.

Frequently Asked Questions

Do I need Data Cloud to install agents from Salesforce AgentExchange? Not for all agents. Base installations of Service Agent and SDR Agent do not require Data Cloud. However, advanced configurations that use cross-channel personalization, unified customer profile data, or next-best-action recommendations do require Data Cloud. Always check the Requirements section of each AgentExchange listing carefully before installing.

Are AgentExchange agents production-ready immediately after installation? No — they're templates, not finished products. What AgentExchange delivers is the agent architecture: the topic structure, sample action references, and starting instructions. Before going live you still need to activate the underlying Flows that actions call, write org-specific instructions, assign the correct permissions to the agent user, connect the agent to a channel, and run end-to-end tests with realistic data.

Is Agentforce available on Salesforce Professional edition? No. Agentforce is only available on Enterprise, Performance, Unlimited, and Developer editions. Professional edition orgs cannot install or use Agentforce agents — this is a hard platform prerequisite that isn't always prominently disclosed in individual AgentExchange listing requirements. Check your edition in Setup → Company Information before investing time in any Agentforce evaluation.

How do you deploy an Agentforce agent from sandbox to production? Do not use Change Sets — Agentforce agent metadata (agent definitions, topics, actions, instructions) does not appear in the standard Change Set component library. The correct approach is a metadata API deployment using Salesforce CLI (sf project deploy start) or by performing a second install directly from AgentExchange into your production org and manually replicating your sandbox configuration. Plan this deployment path before you start configuring in sandbox, not the night before go-live.