The research-to-outreach gap is the most expensive hour in sales
Most reps start the day the same way: open the CRM, scan the territory, try to figure out who to call. That process, pulling account context, finding the right contact, checking for recent news, drafting something worth sending, burns 30 to 45 minutes before a single conversation happens. Multiply that across a team of 20 SDRs and you're looking at hundreds of hours a week spent on work that isn't selling.
The Salesforce Agentforce Prospecting Agent is now generally available, and it changes that math directly. Powered by GTM.AI as the GTM Context Layer, the agent builds a prioritized prospecting queue for every rep before they start the day, with verified ZoomInfo contacts, cited buying signals, and a first-pass outreach draft ready to review. Mutual customers get verified contact, company, intent, and Scoops data inside Agentforce by default: no separate enrichment pass, no CSV import, no broker layer.
The integration runs through Model Context Protocol (MCP), which means any Salesforce customer running Agentforce can wire ZoomInfo's GTM Context Graph directly into the agent's data surface. Early adopter Perk generated 60% of its outbound pipeline in the first two weeks using the Prospecting Agent.
This guide covers what the agent does, how to set it up, what signals it uses, and how the data governance layer works underneath it.
What the Salesforce Agentforce Prospecting Agent actually does
The Salesforce Agentforce Prospecting Agent is an autonomous AI sales agent that runs natively inside Salesforce and builds a prioritized prospecting queue for every rep each day. That distinction matters operationally: this is not a parallel tool with a separate login or a browser extension layered on top of your CRM. It runs inside the Salesforce environment reps already use, against live CRM records, with no additional tool license or integration maintenance.
The agent automates the highest-volume, lowest-judgment portions of a traditional SDR or BDR workflow. Where a Salesforce AI SDR previously meant a rep manually pulling lists, cross-referencing intent data, and drafting cold outreach, the Agentforce Prospecting Agent handles all of that before the rep opens their laptop. Reps start the day reviewing agent output rather than building lists. The agent doesn't speed up prospecting as a rep activity; it eliminates prospecting as a rep activity.
Every row in the daily queue surfaces two contextual columns. "Why this account" explains in plain language why the account is on the list, grounded in concrete signals: BDR team size, YoY growth, technology footprint, CRM activity, funding events, executive moves, and active intent. Each signal cites its source. "Why this prospect" explains why the named contact is the right entry point, combining Salesforce engagement history with GTM.AI signals like job changes, new champions, and intent surges.
The Prospecting Agent operates as one of the sales-focused agents on the Agentforce platform. Agentforce is Salesforce's platform for autonomous AI agents, built on Salesforce MCP to connect AI agents to CRM data. Agentforce agents run across sales, service, marketing, and commerce, inside the Salesforce environment a rep already uses every day. Reps keep ownership of the conversation, the relationship, and the close.
How to set up the Agentforce Prospecting Agent: a step-by-step walkthrough
Setting up the prospecting agent salesforce workflow takes less than a full day for most teams. Here is the sequence:
Step 1: Connect your ZoomInfo entitlement to Agentforce
In the Salesforce setup panel, navigate to the Agentforce data connections configuration and authenticate your ZoomInfo entitlement. Once connected, the Prospecting Agent automatically reads from ZoomInfo's GTM Context Graph as a first-class data source. No manual data exports, no scheduled sync jobs.
Step 2: Define your ICP using the two-tier framework
The agent accepts natural-language ICP input. Use a two-tier structure when defining your target:
Required criteria (hard filters the agent applies to every account): firmographic constraints like company size range, target industries, and geography. These are non-negotiable filters. If an account doesn't meet them, it doesn't appear in the queue.
Nice-to-haves (behavioral and intent signals that boost ranking): funding events in the last 90 days, specific technology in the stack, hiring activity in relevant departments, executive changes in the buying committee.
A practical starting point: define 3 to 5 required criteria and 2 to 3 nice-to-haves. Over-constraining the ICP produces too few results and the queue runs dry. Under-constraining it produces a long list of low-relevance accounts that trains reps to ignore the agent's output.
Step 3: Preview the generated prospect list
Before activating, review the prospect list the agent generates against your ICP definition. This is your quality check: confirm the accounts match your intended profile, spot any obvious mismatches, and adjust the ICP criteria if the list skews in a direction you didn't intend.
Step 4: Activate the agent
Once the preview looks right, activate. Salesforce's own product documentation states that results are visible that same week. The agent begins building daily prioritized queues immediately.
Step 5: Review the daily prioritized queue each morning
Each row in the queue shows the target account, the recommended contact, "Why this account," and "Why this prospect" with each signal cited to its source. Reps can pressure-test the reasoning: if the agent surfaced a funding event as the "why now," the signal traces back to a verifiable source, not a black-box score.
Step 6: Edit, approve, and send the agent-drafted outreach
The agent generates a first-pass outreach message keyed to the signals it surfaced. The rep reviews, edits, and sends. The drafting step uses only what the GTM Context Graph and Salesforce together know about the account and contact. It does not invent facts.
What signals the Prospecting Agent uses to rank accounts and contacts
Understanding the signal layer is what separates reps who trust the agent's prioritization from reps who ignore it. The Agentforce Prospecting Agent draws from two signal categories, and every signal on the queue row is cited to its source. For a broader look at how these signals fit into AI for outbound prospecting, the full guide covers the tools and tactics teams are using in 2026.
GTM.AI Context Graph signals, these come from ZoomInfo's verified data layer and indicate buyer readiness based on external market activity:
Verified contacts: Current title, current employer, verified email, and verified mobile, continuously refreshed. Indicates you're reaching the right person at the right company.
Company firmographics: Revenue range, employee count, industry classification, and growth trajectory. Indicates whether the account fits your ICP at a structural level.
Technographics: Current technology stack, recent additions, and recent removals. Indicates fit with your product's integration requirements or competitive displacement opportunities.
Intent surges: Accounts showing elevated research activity on topics relevant to your category. Indicates active buying consideration, not just passive awareness.
Scoops: Executive moves, funding events, hiring activity in target departments, and technology changes. Each Scoop indicates a specific trigger event that creates a "why now" for outreach.
Hierarchy data: Org chart relationships, subsidiary structures, and parent-child account mappings. Indicates the right entry point within a complex buying organization.
Salesforce CRM signals, these come from your own CRM history and indicate relationship context:
Opportunity history: Prior deals won, lost, or stalled with the account. Indicates whether you're re-entering a known relationship or approaching cold.
Engagement activity: Email opens, call logs, meeting history. Indicates the account's prior responsiveness to your team's outreach.
Objection notes: Logged objections from previous conversations. Indicates what messaging to avoid and what context to acknowledge.
Prior outreach: Sequence history and contact-level engagement. Indicates which contacts have been touched and at what cadence.
Every signal on the queue row cites its source. Reps are not asked to trust a score. They can see exactly what drove the recommendation and decide whether the reasoning holds.
How the ZoomInfo and Salesforce Agentforce integration works
The salesforce agentforce prospecting agent zoominfo integration connects your ZoomInfo entitlement directly to the Agentforce data surface, so every queue the Prospecting Agent builds is grounded in the GTM Context Graph from the moment it activates. There is no separate enrichment pass, no CSV import, and no broker layer between the verified data and the agent.
Account selection
The agent reaches GTM.AI for firmographic, technographic, intent, Scoops, and hierarchy signals, and combines them with Salesforce CRM data to rank accounts by "why now." A rep opening the agent in the morning sees accounts in priority order, with the signal mix that drove the ranking surfaced row by row. The ranking is not a black-box score; it is a cited argument the rep can evaluate.
Contact selection
Once an account is on the queue, the agent uses GTM.AI to identify the right contacts inside it: verified email, verified mobile, current title, current employer, recent job changes, and intent activity. CRM engagement history layers in on top. The rep sees the named contact, the role, the reason that contact is the right entry point, and a draft outreach message keyed to the surfaced signals.
Outreach drafting
The agent generates the first-pass message using the cited signals as the personalization payload. The rep reviews, edits, and sends. The data reaches Agentforce through GTM.AI via API and Model Context Protocol (MCP), the open access lane that exposes ZoomInfo's verified intelligence to any agent or platform that connects to it.
What GTM.AI is and why it matters as the data layer for Agentforce
GTM.AI is ZoomInfo's headless GTM context layer. It exposes ZoomInfo's verified data graph and agentic orchestration through API and Model Context Protocol (MCP), so any tool, agent, or platform can plug in. Salesforce Agentforce is one of dozens of completed integrations on GTM.AI, alongside HubSpot Breeze, Microsoft Copilot Studio, IBM watsonx Orchestrate, Outreach AI, Claude, and ChatGPT. The broader question of how AI agents consume go-to-market intelligence is explored in fueling AI agents with GTM intelligence.
The data quality argument is structural. B2B contact data decays at roughly 70% per year (Salesforce State of Sales). A manual SDR running off a stale list pays for the mistake one bad call at a time. An autonomous AI sales agent running off the same stale data produces thousands of incorrect outreaches a day, erodes brand trust at machine scale, and trains reps to ignore the agent's output. The agent does not get the benefit of human judgment correcting the list as it goes.
ZoomInfo's verified data addresses this directly. The GTM Context Graph holds identity-resolved data on 100M companies, 500M contacts, and 1.5B+ data points processed daily. It fuses ZoomInfo's B2B data with CRM records, conversation intelligence, and behavioral signals into a unified reasoning layer, continuously updated and continuously queryable. Every "why now" the Prospecting Agent surfaces is grounded in real-time GTM intelligence, with each cited signal traceable to a source the rep can pressure-test.
Universal access is the third dimension. GTM Workspace is ZoomInfo's seller-facing product, giving reps direct access to the same GTM Context Graph foundation inside a ZoomInfo-native workspace. For teams building custom agents or integrating into other platforms, the same intelligence is available through APIs and MCP. Same data, same intelligence, no lock-in across any surface a seller or developer needs to reach.
How Agentforce with GTM.AI differs from a generic AI SDR tool
The prospecting agent salesforce architecture differs from a generic AI SDR tool on three dimensions that matter for enterprise buyers.
The first is native CRM placement. The Prospecting Agent runs inside Salesforce against live records, with no separate tool license, no parallel login, and no integration maintenance. A rep does not leave Salesforce to use the agent. That sounds like a convenience feature; it is actually a data integrity feature. When the agent reads from live CRM records rather than a synced copy, the prioritization reflects the current state of every account, including deals in flight, recent objections, and prior contact history.
The second is the verified data layer. Most AI SDR tools answer against whatever data the customer brings. With GTM.AI as the GTM Context Layer, the agent reads from the GTM Context Graph by default: 500M verified contacts, 120M direct-dial phone numbers, 200M+ verified business emails, continuously refreshed and identity-resolved. The data underneath the agent is not a static import; it is a live, continuously updated view of the market.
The third is cited reasoning with platform-level governance. The agent surfaces the signals that drove each recommendation. GTM.AI's platform layer applies access control, permissioning, data lineage, AI policy, and audit logging consistently across every agent surface. Every agent output is traceable to a source. Reps see why an account is on the list. Compliance teams see a governed, auditable AI layer. That combination, cited reasoning plus platform-level governance, is the structural difference between an agent that enterprise buyers can deploy and one they cannot.
What sales leaders and reps can expect from day one
Industry benchmarks across the AI sales agent category show that what used to take a rep 45 minutes of manual research can now take seconds when the agent has verified, continuously refreshed data underneath it. The gap between research and outreach collapses, and that time goes back to selling.
Early adopter Perk reports that the Prospecting Agent generated 60% of its outbound pipeline in the first two weeks of use. The metric measures pipeline, not activity or meetings booked. It is a working signal that an AI sales agent grounded in verified data produces real revenue, not vanity output.
The second named customer outcome comes from a team that has been running on ZoomInfo's GTM intelligence layer longer. Seismic saved 11.5 hours per rep each week and attributed 39% of pipeline to ZoomInfo signals. Eleven and a half hours a week is not a productivity rounding error; it is more than a full selling day returned to quota-carrying work.
The structural shift underneath both outcomes is the same. Pipeline ownership becomes co-production: the agent owns prioritization and drafting, the rep owns conversation, relationship, and close. The rep's job does not disappear; it moves up the value chain. Every CRM platform is shipping an agent layer. The differentiation comes from the freshness, connectivity, and verification of the data underneath that agent, and from a single governance posture across every agent in the stack.
Data governance and compliance: how the agent handles accuracy and privacy
For enterprise buyers, the governance question comes before the capability question. An AI agent that produces unauditable output, or that surfaces data without a clear lineage, is a procurement blocker regardless of how good the prioritization logic is.
GTM.AI's governance plane applies access control, permissioning, data lineage, AI policy, and audit logging consistently across every surface that consumes it. Agentforce, the Prospecting Agent, and any future agent the customer builds inherit the same governance posture. There is no separate governance configuration for the Agentforce integration; the platform-level governance applies automatically. Every agent output is traceable, every data access is logged, and every policy set at the GTM.AI level propagates to every connected surface.
Hallucination risk is addressed structurally. The Prospecting Agent does not invent facts. Every outreach draft uses only what the GTM Context Graph and Salesforce together know about the account and contact, with each signal cited to its source. The agent cannot surface a claim it cannot trace to a verified data point. That is not a guardrail bolted on after the fact; it is how the data layer is architected.
ZoomInfo's compliance credentials cover the enterprise procurement checklist: ISO 27001, ISO 27701, SOC 2 Type II, and TRUSTe GDPR/CCPA certification. These apply to the full GTM.AI platform, including the Agentforce integration.
For teams with EU-specific data compliance requirements, confirm coverage details with your ZoomInfo account team. EU data coverage specifics are available on request, and your account team can walk through the relevant certifications and data residency options for your geography.
To see how the Prospecting Agent handles data governance for your team, request a demo.
Availability and how to get started
The Salesforce Agentforce Prospecting Agent is generally available today to Salesforce customers running Agentforce, with ZoomInfo's GTM.AI wired in as the GTM Context Layer. Mutual customers can connect their ZoomInfo entitlement to Agentforce and see results that same week.
For teams that want the same verified intelligence in a ZoomInfo-native seller workspace, GTM Workspace provides the same GTM Context Graph foundation in a seller-facing product built for daily prospecting workflows. GTM Workspace and the Agentforce integration draw from the same data layer, so teams running both get consistent intelligence across every surface their reps work in.
To connect your ZoomInfo entitlement to Agentforce and see the Prospecting Agent in action, request a demo.
FAQ: Salesforce Agentforce Prospecting Agent and the ZoomInfo integration
What is the Salesforce Agentforce Prospecting Agent?
The Salesforce Agentforce Prospecting Agent is an autonomous AI sales agent that runs inside Salesforce and builds a prioritized prospecting queue for every rep each day. It identifies target accounts, surfaces the right contacts with verified data from ZoomInfo's GTM Context Graph, explains the timing rationale for each recommendation, and drafts personalized outreach before the rep starts the day. Reps keep ownership of the conversation and the close. The salesforce prospecting agent is generally available now to Salesforce customers running Agentforce.
How does the ZoomInfo and Salesforce Agentforce integration work?
Mutual customers connect their ZoomInfo entitlement to Agentforce, and from that point the agentforce prospecting agent reads verified contact, company, intent, Scoops, and technographic data from ZoomInfo's GTM Context Graph as a first-class data source. No separate enrichment pass, no CSV import. The integration runs through GTM.AI, ZoomInfo's headless GTM context layer, which exposes the data graph through API and MCP.
What is GTM.AI?
GTM.AI is ZoomInfo's headless GTM context layer. It exposes ZoomInfo's verified data graph (500M contacts, 100M companies, billions of signals), agentic orchestration, and platform-level governance through API and Model Context Protocol, so any agent, platform, or workflow can plug in. Salesforce Agentforce is one of dozens of completed integrations on GTM.AI, alongside HubSpot Breeze, Microsoft Copilot Studio, IBM watsonx Orchestrate, Outreach AI, Claude, and ChatGPT.
What results have early customers reported?
Early adopter Perk reports that the Prospecting Agent generated 60% of its outbound pipeline in the first two weeks of use. The metric measures pipeline, not activity or meetings booked, which makes it a meaningful signal that an AI sales agent grounded in verified data produces real revenue. See Seismic's pipeline results: their sales team saved 11.5 hours per week per rep and attributed 39% of pipeline to ZoomInfo signals.
Does the Prospecting Agent use MCP?
GTM.AI exposes its capabilities through Model Context Protocol in addition to a standard REST API. Agentforce agents reach GTM.AI through the surface that fits their workflow. The same MCP access lane that powers the Agentforce integration also connects GTM.AI to Claude, ChatGPT, and other AI agent platforms.
Is the Salesforce Agentforce Prospecting Agent available now?
Yes. The salesforce prospecting agent is generally available to Salesforce customers running Agentforce, with ZoomInfo's GTM.AI wired in as the GTM Context Layer by default. Mutual customers can connect their ZoomInfo entitlement to Agentforce and see results that same week. Request a demo to get started.