Prompts

Everything the exercises ask you to type or paste, in one place. It's grouped by section and exercise, in the same order as the guide. To copy a block, hover over it and select the copy icon on the right.

Agentforce Prospecting Agent

Exercise 0 - Set Up Your Sandbox

Step 6: Create a Test Account. Example values. Use another real company if it fits your ICP better.

Field	Value
Account Name	Snowflake
Industry	Technology
Employees	7,800
Website	snowflake.com
Billing Country	United States

Step 7: Add a Test Contact. Set Title to:

VP of Sales

For Email, use your own address.

Exercise 1 - Turn On Prospecting and Set Up Access

Step 5: Add Prospecting to the Navigation Bar. If Prospecting isn't in the nav bar yet, add this to the end of your org's URL:

/lightning/page/prospecting

Exercise 2 - Create Your Prospecting Agent

Step 2: Name Your Agent. Enter an Agent Name (Internal), for example:

Prospecting Agent - NA Enterprise

Exercise 3 - Teach the Agent Your ICP

These criteria are examples. Replace them with your own, based on your ICP answers from Exercise 0.

Step 1: Define Your Target Accounts (Account Research). Each block is one bullet. Click a bullet, press Cmd+A (Ctrl+A on Windows), and paste. To add a bullet, press Enter at the end of the last one.

Required Criteria:

Use Account.Industry. Include accounts where Industry is Technology, Software, or Telecommunications. If Industry is blank, use web search to find the industry. If excluded, give the reason: "Industry is outside ICP."

Use Account.NumberOfEmployees. Include accounts with more than 500 employees. If NumberOfEmployees is blank, use web search to find the company size, and apply the same threshold. If excluded, give the reason: "Company size below ICP threshold."

Use Account.BillingCountry. Include only accounts in the United States. If excluded, give the reason: "Geography outside ICP."

Optional Criteria:

Rank accounts higher when Account.LastActivityDate is within the last 90 days. The more recent the activity, the higher the rank. Do not disqualify accounts with older or blank activity dates.

Rank accounts higher in industries next to our ICP, such as FinTech or Cybersecurity, when Account.AnnualRevenue is greater than $100M.

To have an AI assistant, such as Slackbot, draft criteria for you, paste your ICP notes along with this pattern:

Where to look (field or source) → the rule (include, exclude, or rank) → what to do if data is missing → the reason the agent should give

Step 3: Define Your Target Contacts (Find Prospects). Each section is its own box.

Ideal Prospect:

Director level or above, with a title containing Sales, Revenue, Marketing, or Operations. Prefer contacts with a verified email address.

Acceptable Prospect:

Managers in Sales Operations, Revenue Operations, or Marketing Operations.

Poor Prospect:

Executive Assistants, Administrative Coordinators, consultants, board members, interns, and anyone marked Do Not Contact.

Exercise 4 - Activate and Test Your Agent

Step 1: Set Up Assignment for Testing. Set open prospects per sales rep to 10.

Step 3, Option 2 (requires ZoomInfo). For Maximum Number of Prospects, enter 1.

Agentforce Engagement Agent

Exercise 1 - Set up Engagement Agent

Step 2: Enable Agent to Send Emails and Book Meetings. Name for the Einstein Activity Capture configuration:

EAC Config

The agent's user record can have any first and last name. The guide uses:

Field	Value
First Name	Welo
Last Name	Agent

Step 3: Configure Data for your Agent (Optional). Data library name:

Welo Technologies Sales Data Library

Description:

PDF files containing information about Welo Technologies, FAQ's, Pricing and anything Sales related.

File to upload, if you don't have your own: Welo Technologies PDF

Exercise 2 - Build Engagement Agent

Step 1: Build your Agent with Guided Configuration. Use your own company's details, or the values shown in the guide's screenshot.

Agent Name (Internal):

Welo Engagement Agent

Company Name:

Welo

Company Description:

Welo is a cutting-edge software company specializing in decentralized workplace collaboration tools and AI-driven productivity ecosystems.

Primary Value Proposition:

Empower teams to move from static task management to dynamic, automated work execution—driving up to 2,000+ hours saved annually through real-time enterprise visibility, seamless cross-functional alignment, and no-code automated workflows.

Key Achievements:

Operational Efficiency: Delivers a 25% average gain in team productivity and eliminates manual status reporting by consolidating portfolio data into real-time dashboards.
Enterprise Security & Scale: Trusted by 90% of the Fortune 100, offering enterprise-grade governance, SAML/SSO, and fine-grained access control.
Accelerated Time-to-Value: Speeds up project delivery timelines by 30% via automated review processes, integrated approval paths, and core integrations (Salesforce, Jira, Slack).
ROI Impact: Drives a 380% Forrester-verified ROI over three years, with payback realized in under 6 months.

Sample customer message for the Response Email preview:

How does pricing work?

Agentforce Coworker

Exercise 1 - Setup

Step 2: Ask Agentforce Coworker a Question.

What deals should i focus on for the remainder of the year?

If Coworker doesn't return any deals, try:

What are my 5 largest open opportunities and what should I do next on each?

Exercise 2 - Connecting Agentforce Coworker to a MCP Server (Optional)

Step 1: Register a new MCP Server.

MCP Server Name:

PubCrawl

Description:

MCP Server to provide access to biomedical literature and drug data sources

Server URL:

https://pubcrawl-production-feb9.up.railway.app/mcp

Set Authentication Method to No Authentication. It defaults to OAuth 2.0.

Trusted URLs, one for each row. Leave the other settings as they are.

API Name	URL
DOI	doi.org
PubMed	*.ncbi.nlm.nih.gov

Step 2: Create a new Employee Agent.

Agent Name:

PubCrawl Employee Assistant

Description, in Agent Settings:

Search peer-reviewed medical literature, clinical trials, patents, pharma articles, and pharmaceutical drug labeling across PubMed & PubMed Central, Europe PMC, ClinicalTrials.gov, Drug Labeling (US FDA & UK eMC).

Subagent Name:

PubCrawl Scan

Subagent Description:

Search peer-reviewed medical literature, clinical trials, patents, pharma articles, and pharmaceutical drug labeling across PubMed & PubMed Central, Europe PMC, ClinicalTrials.gov, Drug Labeling (US FDA & UK eMC).

Utterance for Preview:

What articles about CAR-T Therapy were published in 2016?

Step 4: Confirm that the agent is visible to your user. Sample utterance:

What articles about CAR-T Therapy were published in 2016?

Step 5: Use Coworker with MCP connection enabled. Ask Coworker:

What articles about CAR-T Therapy were published in 2016?

For the meeting brief, Log a Call on a lead (for example, Katherine Slayden) with this comment:

Working on a new drug for CAR-T, currently in clinical trials.
Trying to improve outcomes in patients with the CD20 and CD22 biomarkers.
Reminder: research clinical research history for CAR-T CD20/CD22 before next mtg

Then ask Coworker (change the name if you used a different lead):

Brief me for my upcoming meeting with Katherine Slayden

If Coworker offers to research CAR-T CD20/CD22, reply:

Yes

If the brief doesn't mention the call you logged, try:

Pull up the recent activity on Katherine Slayden and research it.