AI is already showing up in the day-to-day work of revenue teams. It summarizes calls, drafts follow-up emails, researches accounts, updates CRM records, and points out deals that may need attention.

Some of it is genuinely helpful. Some of it creates more work than it saves.

That is the reality of AI for RevOps. The technology can make a good revenue process faster and easier to manage. It can also take messy data, unclear rules, and bad handoffs and spread those problems across the business.

The real question is not whether a company should use AI. It is where AI can help today, where the risk is still too high, and what needs to be fixed before giving it more responsibility.

AI Is Good at Clearing Away Busywork

Revenue teams spend a surprising amount of time on work that is necessary but not especially valuable.

Sales representatives write call notes, update next steps, research accounts, and prepare for pipeline reviews. RevOps teams chase down missing fields, clean up records, and answer questions that could have been handled through a report or dashboard.

AI can take a meaningful amount of that work off the team’s plate.

A call summary does not need to be perfect on the first try. It needs to give the seller a useful starting point. The same is true for a follow-up email, account brief, or suggested task. A person can review the output, correct anything that is off, and move on faster than if the work had started from a blank page.

Salesforce describes Agentforce as an AI agent platform that can answer questions, retrieve company knowledge, and complete approved actions. For many RevOps teams, the safest place to start is with tasks that save time but still leave the final decision with an employee.

That may not sound as exciting as an agent running an entire process on its own. It is often much more useful.

CRM Maintenance Is an Obvious Opportunity

Nobody goes into sales because they enjoy updating opportunity fields.

That creates a familiar problem. Representatives wait until the forecast meeting to update Salesforce. Managers question the numbers. RevOps sends reminders. Everyone spends Friday morning trying to figure out what actually changed during the week.

AI can help by suggesting updates from calls, emails, and recent activity. It can identify missing next steps, flag old close dates, or point out opportunities that have not moved in weeks.

The key word is “suggesting.”

An AI-generated recommendation can save a representative time. An automatic update to a forecast category can affect reporting, hiring plans, and leadership decisions. Those two actions should not have the same level of oversight.

Low-risk work can move quickly. Anything that changes financial reporting, pricing, contracts, or customer commitments needs more control.

AI Can Find Patterns People Miss

Revenue teams have more data than most employees can reasonably review.

A sales manager may have dozens of open opportunities. Marketing may be tracking thousands of leads and campaign responses. Customer success teams may be watching usage, cases, renewals, and health scores across an entire customer base.

AI can sort through that activity and highlight where someone should look.

It may notice that a large opportunity has gone quiet. It might identify accounts with falling product usage or leads that resemble past customers. It can also find records with unusual activity that a weekly dashboard would not surface.

That is valuable, but it is not the same as knowing what should happen next.

An opportunity may look inactive because an important conversation happened outside Salesforce. A customer with lower usage may be experiencing a seasonal dip rather than a renewal risk. A lead may resemble a high-value account on paper but still be a poor fit.

AI is good at pointing to the smoke. Someone who understands the business still needs to check for the fire.

Poor Data Does Not Become Better Because AI Uses It

This is where the excitement around AI tends to run into the reality of CRM data.

If Salesforce contains duplicate accounts, missing contacts, stale opportunities, and inconsistent field values, the AI has an incomplete view of the business. Its answer may sound polished while still being based on the wrong information.

Consider an agent that recommends which deals need attention. It may look at stage, close date, activity, next steps, and contact engagement. If representatives rarely update those fields, the recommendation will not be very useful.

Connecting more systems does not automatically solve the problem either. Data 360, formerly Data Cloud, can bring together information from Salesforce and other platforms. However, teams still need to decide which source owns each data point, how records should be matched, and what information the AI can trust.

This is one reason many companies are not ready for Agentforce yet. They may be ready to test a focused use case, but not to let AI operate across an entire revenue process.

That is not a reason to stop. It is a reason to narrow the scope.

Automating a Broken Process Just Makes It Break Faster

Every revenue organization has at least one process that works mainly because employees know how to work around it.

Lead routing is a common example. Marketing and sales may use different definitions of a qualified lead. Territory rules may be outdated. Ownership exceptions may live in someone’s spreadsheet.

Adding AI to that process will not settle those disagreements. It will simply make decisions faster using whatever rules and data it receives.

Before automating anything, the team needs to walk through what actually happens today. Where does the process begin? Who makes each decision? What happens when the usual rules do not apply? Which team owns the outcome?

If those answers change depending on who is asked, the process needs attention before the AI does.

The strongest AI use cases usually come from processes that are already understood. AI removes unnecessary steps, finds information faster, or helps employees handle volume. It should not be expected to repair years of operational confusion on its own.

Customer-Facing AI Carries More Risk

There is a big difference between an AI tool drafting an internal account summary and an agent speaking directly to a customer.

If an internal summary contains a mistake, an employee has a chance to catch it. If a customer-facing agent gives the wrong answer about pricing, contract terms, or account access, the damage has already started.

That does not mean customer-facing AI should be avoided. It means the boundaries need to be clear.

An agent may be able to answer common questions, check an order status, schedule an appointment, or direct a customer to approved information. Sensitive complaints, unusual requests, cancellations, and contract questions may still need a person.

Salesforce’s responsible agentic AI guidelines emphasize accuracy, safety, transparency, and effective handoffs between AI and people. Those ideas become especially important when the AI represents the company directly.

A customer should not have to argue with an agent to reach a human.

RevOps Has a Role in AI Governance

AI governance can sound like something legal, security, or IT should handle. Those teams need to be involved, but they do not own the revenue process.

RevOps knows how leads move between teams, how opportunity data affects the forecast, and where automation touches the customer journey. That perspective is necessary when deciding what an AI agent should be allowed to see and do.

Someone needs to own each use case. Someone also needs to approve changes, review results, and respond when the AI makes a poor recommendation.

Access matters too. An agent should not receive broad permissions simply because that makes the build easier. It should have access to the information and actions required for its job—and nothing more.

Salesforce’s Data 360 governance tools can support access policies, data classification, masking, and encryption. The technology can enforce a policy, but the business still has to decide what that policy should be.

The Revenue Ops guide to AI governance for revenue teams covers the practical questions that need answers before AI moves into production.

Start With a Job, Not an AI Strategy

“Use AI to improve sales productivity” sounds like a goal, but it does not tell anyone what to build.

“Create a draft call summary and suggest the next step” is much clearer.

A focused use case has a starting point, an expected result, and a way to measure whether the tool helped. The team can compare how long the task took before and after AI. Users can review the quality of the output. RevOps can see whether the information in Salesforce improved.

That makes it easier to find problems before the use case expands.

If the AI works consistently, give it another related task. If employees keep correcting the same issue, fix the data, instructions, or process before moving forward.

There is no prize for launching the most AI use cases at once. The goal is to introduce something the team will trust and continue using after the initial excitement wears off.

The Right Amount of AI Will Be Different for Every Team

AI for RevOps has a useful role in research, summaries, CRM maintenance, analysis, and routine automation. It can give employees time back and help them find information that would otherwise remain buried.

The risk grows when AI works with unreliable data, follows an unclear process, or takes action without the right checks in place.

For some teams, the next step may be a focused Agentforce pilot. For others, it may be cleaning Salesforce data, simplifying a workflow, or deciding who owns AI governance. That foundational work is not separate from the AI strategy. It is what makes the strategy realistic.

Start with a problem worth solving. Keep the first use case manageable. Let employees challenge the output. Then expand only when the results show that the business—and the AI—are ready.