100 Ready-to-Use AI Prompts for Salesforce Admins

100 Ready-to-Use AI Prompts for Salesforce Admins

By

In this blog post, we looked at how the writing skills you already use every day carry over to prompting an agent. Now let’s put them to work even more. At our Dreamforce 2026 session on Agentforce Vibes, Lay Durafe and I handed out something to practice on: 100 ready-made prompts for the jobs admins do every week.

Lay built the Agentforce Vibes Admin Accelerator Kit and hosts it on GitHub. Don’t worry about the GitHub part. You won’t need to clone a repo. GitHub is a handy place to host the kit’s markdown file, and you can read and copy every prompt right in your browser. The kit sorts its prompts into four buckets:

  1. Understanding
  2. Configuring
  3. Troubleshooting
  4. Documenting

Each bucket holds five categories with five prompts apiece. Anything in angle brackets is a placeholder, so swap <Object>, <Field_Name> or <Flow_Name> for the actual API names in your org. Our demo used Agentforce Vibes, and the prompts also work with other agents connected to a Salesforce org through MCP. Start in a sandbox or Developer Edition org, especially with Configuring prompts. Understanding and Documenting prompts mostly read your org’s metadata. Configuring prompts change it.

Here’s how to put each bucket to use.

Understanding: get to know your org first

You don’t need a plan mapped out to get value from an agent. Exploring your org with one saves you time and shows you what’s actually happening in your org.

The Understanding bucket covers objects, Flows, permissions and business processes. A few to start with:

  1. “Explain what the Flow <Flow_Name> does in simple terms.”
  2. “List all active Flows running on <Object>.”
  3. “Map out the full lifecycle of a <Object> record.”
  4. “Why can <User> edit <Field_Name> but <User_2> cannot?”

Check what breaks before you change it

Picture this. Sales wants the free-text Region field on Account turned into a picklist, so reports stop showing “NE,” “Northeast” and “north east” as three different places. Before you touch it, ask:

  1. “If I change <Field_Name> from text to picklist, what is impacted?”
  2. “List all Apex, Flows, and reports using <Field_Name>.”
  3. “Which page layouts and permission sets include <Field_Name>?”

Five minutes of questions here beats hearing about the integration that writes to that field from an angry email on Monday morning.

Configuring: ask for directions or hand over the keys

Map out your project with “How do I” prompts

Prompts that start with “How” or “Why” help you understand the process behind a task. “How do I set up a lookup relationship between <Object_A> and <Object_B>?” gets you a list of Setup paths and clicks. That might be all the documentation you need to do the task yourself. And if the agent can do the work, its answer shows you exactly what it plans to do before you send it off to do it.

Precise prompts can still start big. Ask about the whole project first, then break it down. Try “What would you create to <goal>? Don’t build anything yet.” You get a starting point to talk through the plan, turn it into requirements and start building.

Need a project plan too? Ask the agent to write that up as well.

Get it done with “Create” and “Write” prompts

Start a prompt with a strong action verb and expect action back. The agent still won’t change your org without your permission. “Create a Flow that updates <Field_A> to <Value> when <Field_B> changes” generates the metadata you need to deploy the Flow. That metadata waits for your review, and deploying it to the org takes a separate prompt.

When you’re ready to hand over the keys and let the agent create or edit parts of your org, precise prompts get you there fast:

  1. “Build a Record-Triggered Flow on <Object> to create a child <Object_2> record.”
  2. “How do I add a scheduled path to a Flow for <X_Days> after record creation?”
  3. “Write a formula to require <Field_A> only when <Field_B> equals <Value>.”
  4. “How do I prevent users from editing <Field_Name> after stage <Stage_Name>?”
  5. “Create a validation rule that ensures <Field_Name> is formatted correctly.”

Some changes fall outside what the agent can do, and it will tell you when you hit one. For those, ask how to do it yourself or just ask for advice:

  1. “How do I add a field to the record page layout for <Object>?”
  2. “How do I conditionally hide a component on the <Object> page for <Role>?”
  3. “Suggest a simple layout design for <Object> for <Team_Name> users.”

Troubleshooting: let the agent do the digging

When you aren’t exploring your org, you’re probably figuring out what broke in it. An agent gives you quick answers to questions like:

  1. “Fix this Flow error: <Paste_Error>.”
  2. “Why is <Flow_Name> failing at element <Element_Name>?”
  3. “Why is <Flow_Name> throwing an unhandled fault error on <Object>?”
  4. “Why is <User> getting an ‘Insufficient Privileges’ error on <Action>?”
  5. “Troubleshoot why <User> cannot see the <Tab_Name> tab.”
  6. “Why can’t <Profile> create new <Object> records?”

One tip: paste the whole error. “Fix this Flow error: <Paste_Error>” works best with the full fault email, plus the record ID and the user who hit it. The more evidence you hand over, the less the agent has to guess.

When automations fight each other

Some tickets come with no error at all. A rep saves a record, the Status field flips back to its old value, and nobody knows why. The kit has a set of prompts for exactly this:

  1. “Which automations are changing <Field_Name> on <Object>?”
  2. “Why is a record saving, but fields are changing back automatically?”
  3. “Check if <Flow_A> and <Flow_B> on <Object> conflict with each other.”

Tracking that down by hand means opening every Flow on the object and reading each one. An agent connected to your org can scan them all and tell you where to look first.

Documenting: write it down while it’s fresh

Your conversations with an agent can outlive the task. Document as you go:

  1. “Create a data dictionary table for <Object> listing field names and types.”
  2. “Document all custom fields created on <Object> and their business purpose.”
  3. “Summarize all object relationships connected to <Object>.”

When it’s time to ship, the kit has prompts that cover the deployment and the paperwork around it:

  1. “Write release notes for these updates: <List_Changes>.”
  2. “Create a deployment checklist for moving <Flow/Fields> to Production.”
  3. “Draft communication to users about upcoming changes to <Object/Process>.”

Capture the why

If I had to point you to one prompt in this bucket, it’s “Document why <Validation_Rule/Flow> was configured this way.” Every admin has inherited a rule with no description and nobody left to ask about it. Run this prompt right after a build, while the reasons are still sitting in your conversation with the agent. The next admin gets the answer for free, and that admin might be you in two years.

Putting the buckets together

Each bucket works on its own, and they also stack. Back to that Region picklist:

  1. Understanding: “If I change <Field_Name> from text to picklist, what is impacted?”
  2. Configuring: “How do I create a picklist field on <Object> with values <Value_1, Value_2>?” Or ask the agent to build it.
  3. Troubleshooting: if a Flow starts failing after the change, “Why is <Flow_Name> failing at element <Element_Name>?”
  4. Documenting: “Write release notes for these updates: <List_Changes>.”

Each answer feeds the next prompt. That’s the follow-up habit from the last post, with the kit giving you a starting point at every step.

Tell us what’s missing from the kit

Lay and I want the kit to keep growing with the prompts that work for you. Tell us which ones helped and where you found gaps. If you know your way around GitHub pull requests, open one on Lay’s repo or give Lay a ping on LinkedIn.

Pick one bucket this week, run three of its prompts against a sandbox, and see what you learn about your org.

Resources

8 Lessons for Writing AI Prompts as a Salesforce Admin

8 Lessons for Writing AI Prompts as a Salesforce Admin

Agents are turning into everyday assistants for your work, and AIforce gives those assistants a direct line into your Salesforce org. At Dreamforce 2026, my colleague Lay Durafe and I showed how Salesforce admins can use the same tools developers use to vibe code, and vibe admin instead. The more you work with agents, the […]

READ MORE
introduction to AIforce in Action

Introduction to AIforce for Admins

In September, Salesforce announced AIforce, its strategy for bringing Salesforce to any AI model, agent, or interface where your teams already work. To see where it fits, picture the Salesforce platform as four layers. Data 360 brings your data together. Customer 360 holds your workflows, data models, and business logic. Agentforce is where agents get […]

READ MORE
Streamline Your Path to Production with Builder Central

Streamline Your Path to Production with Builder Central

Admins are managing more requests than ever, and let’s be honest, it’s getting harder to keep pace. Turning business ideas into production-ready apps and agents takes time. You’re constantly stitching together features, navigating complex data models, and toggling between Setup and different builders. At the end of the day, it means that you can’t keep […]

READ MORE