AI, Salesforce, Scaling Operations

AI Speeds Up Salesforce Work. Experience Keeps It Safe.

AUTHOR

Brian Knapp

Published

November 17, 2026

Read Time

4 min read

At Dreamforce, AI was in nearly every keynote, session, and booth conversation. I came home convinced of two things. AI is changing how Salesforce work gets done. And the teams that hand it the keys without guardrails are going to learn some expensive lessons.

What AI Has Changed in My Daily Work

I use Claude every day. It has changed how I spend my time more than any tool I’ve picked up in my career. I can create a build, test it, and revise it in a fraction of the time it used to take.

Here’s a recent example. I updated a screen flow: added several new screens, built new logic for assigning record ownership, and removed an email notification. Then I tested it and wrote the test scripts. That work would normally take me a few hours. With Claude, I finished all of it, testing and scripts included, in about an hour.

But Claude didn’t design that change. I did. I gave it the exact screens to build and the fields on each one, the logic for assignment and updates, and the specific request to remove the email. Claude handled the build. The decisions came from me.

That speed is real. But speed was never the hard part of Salesforce work.

The Hard Part Is Knowing What to Ask

I’ve spent more than 10 years in the Salesforce ecosystem. Before that, I built my own lightweight CRM in Excel. It was crude, but it taught me how records relate to each other, why data structure matters, and what breaks when you get it wrong. That foundation is what makes AI useful to me instead of risky.

AI answers the question you asked. It doesn’t tell you that you asked the wrong one. Experience shows up in three places:

  • What to ask. A good prompt includes the context AI can’t see: your data model, existing automation, sharing rules, and how the business actually uses the org.
  • What it gives back. Output can look clean and still be wrong. You need to be able to read it and spot what’s missing.
  • Whether it’s best practice. AI will happily produce something that works in a demo and fails in production: logic that isn’t bulk-ified, hardcoded IDs, automation that collides with an existing flow, or code that ignores field-level security.

Here’s one from my own work. Claude wrote an automated morning update for an internal Slack channel. It looked right, but when I tested it through Slack, the message showed raw HTML markup instead of formatted text. The content was fine; it was built for the wrong place.

I caught it because I tested it where people would actually read it, so it never reached anyone. The fix was logic that formats the message based on where it’s sent.

Without that experience, you don’t know what you don’t know, and AI won’t tell you.

“AI answers the question you asked. It doesn’t tell you that you asked the wrong one.”

Without Guardrails, AI Makes Changes You Didn't Intend

AI tools that can act inside your org don’t just suggest changes. They make them, and they make them fast. A misunderstood instruction can update the wrong records, overwrite a field across thousands of rows, or change metadata that other automation depends on. A human making the same mistake usually catches it after a few records. AI can finish the job before anyone notices.

That’s not a reason to avoid AI. It’s a reason to set rules before you use it.

The Guardrails I'd Put in Place First

  • Work in a sandbox first. Every time, no exceptions for “quick” changes.
  • Have a backup plan for every change, even the smallest one. Export the data before a data change. Keep metadata in source control. Know exactly how you’d roll back before you start.
  • Limit who can use AI against your org. Not everyone with a Salesforce login should be able to point AI at production.
  • Limit what the AI can touch. Run it under a user with only the permissions the task needs. If the AI can’t edit a field, it can’t break it.
  • Put a qualified human between AI and production. Someone who knows the org reviews every AI-built change before it deploys.

The Takeaway

AI is a multiplier. In experienced hands, it multiplies output. Without experience and guardrails, it multiplies mistakes. The Salesforce teams that win with AI won’t be the ones who adopt it fastest. They’ll be the ones who pair it with people who know what good looks like.

Bringing AI Into Your Salesforce Org?
We help teams put the right guardrails, permissions, and review process in place so AI speeds up the work without putting the org at risk.

AI, Salesforce, Scaling Operations

Share