When the AI conversation started gaining momentum a couple of years ago, I remember sitting in meetings where companies were eager to jump on the trend. Everyone wanted an AI solution, whether it was for their own operations or something they could offer their clients.

Some of those ideas had potential. Others felt half-baked.

I raised my skepticism in those meetings, not because I didn’t believe AI could be useful, but because I wasn’t convinced every problem needed an AI solution in the first place.

And I still feel that way.

Not every process needs an AI agent. Not every business challenge requires a new tool. Sometimes, the answer is simply fixing a process that wasn’t working properly to begin with.

I also think we’ve been looking at AI adoption the wrong way. Adoption should not be another KPI that organizations need to hit just to show progress. It’s a change management challenge. Employees need to understand why a tool matters, how it fits into their work, and whether it actually makes their jobs easier.

That’s why the conversation around Slackbot and Agentforce interests me.

Salesforce has spent years encouraging organizations to put more of their customer data and processes into its ecosystem. Meanwhile, Slack has become a place where employees discuss opportunities, share updates, and coordinate the work that follows.

With Slackbot becoming more capable and Agentforce expanding across Salesforce, the place where people work and the place where business processes run are starting to overlap.

That sounds promising. But it also raises a question: Just because employees can do something in Slack, does that mean the process should live there?

Slackbot Has Become More Than a Slack Assistant

For a long time, Slackbot was associated with reminders, notifications, and simple automated responses. That’s no longer an accurate picture of the product.

In 2026, Salesforce rolled out a much more powerful Slackbot that understands the context of the workplace and helps people work across conversations, documents and connected apps.

According to Slack’s announcement of its context-aware AI agent, Slackbot can use information that employees are authorized to access, including messages, channels, files, and connected tools.

Think about someone joining a client project halfway through implementation. There may be weeks of discussions, documents, unresolved issues, and decisions buried in different channels. Getting that person up to speed usually involves reading through conversations, asking questions, and hoping someone remembers why a particular decision was made.

Slackbot can help pull that context together.

It can also support research, document creation, and work across connected applications. With its expanded capabilities, including deeper research and reusable skill sets, Slackbot is becoming more than a tool for finding information.

I can see why that would appeal to organizations where Slack is already central to day-to-day operations.

There is one thing I would be careful about, though.

Having access to more information doesn’t necessarily mean the assistant knows which information should take precedence.

A conversation from last Tuesday might explain why someone wanted to change a process. That doesn’t mean the change was approved or implemented.

Those distinctions matter once AI starts influencing actual business decisions.

Agentforce Approaches the Problem From the Business Process Side

Agentforce is not designed and deployed the way organizations usually do.

Slackbot starts from the work context of the employee, but Agentforce can be set up around specific business responsibilities, data sources, instructions and actions.

Let’s say your organization wants to improve how you manage stalled opportunities.

An AI agent can review a Salesforce record, find deals that match the inactivity criteria, and recommend follow-up actions. Based on the agent’s configuration, it can also create tasks or update records using approved actions.

That sounds straightforward until you start asking what stalled actually means.

Is it an opportunity with no activity for 14 days? Does that apply to every sales stage? What happens when the customer has already scheduled a meeting for next month? Should a strategic account follow the same rules as a smaller opportunity?

Someone has to make those decisions.

Agentforce gives organizations a way to configure AI around defined processes, but the quality of the outcome still depends on the quality of those processes.

I’ve made a similar argument in our article on what organizations should do before implementing Agentforce.

The technology can help execute the work. It shouldn’t be responsible for figuring out how your business is supposed to operate.

And that’s an important distinction when companies are trying to justify their AI investments.

Before You Implement Agentforce, Look at Your Foundation

If your Salesforce environment has inconsistent data, unclear ownership, or processes that vary from team to team, those problems deserve attention before you introduce more automation.

Slackbot vs. Agentforce: What Should Each One Handle?

I wouldn’t approach this as a traditional product comparison where one platform wins based on the number of features it offers.

The capabilities are increasingly overlapping.

Slackbot can work across connected applications and coordinate agents. Agentforce can operate through Slack. Salesforce has also introduced more ways for employees to interact with CRM information directly from conversations.

So the better distinction is the kind of responsibility each system is taking on.

These are practical distinctions rather than hard technical limitations. Slackbot can also perform actions, including supported Salesforce record operations, while Agentforce can provide information and assistance without changing records.

The question is which approach makes the most sense for the process you’re trying to improve.

The Real Opportunity Is Connecting the Conversation to the Action

Here’s an example.

A sales manager opens Slack on Monday morning and asks which opportunities need attention before the weekly pipeline meeting.

There are two parts to that request.

First, the manager needs context. What happened last week? Were there customer objections? Did someone mention a delay? Has the team already agreed on next steps?

Much of that information might be sitting in Slack conversations.

Then there’s the operational side. Which opportunities have changed stages? Which close dates moved? Are there records with missing next steps? Which deals meet the company’s criteria for escalation?

That’s where structured Salesforce data becomes important.

In a connected implementation, Slackbot could help the manager interact with the information while a configured Agentforce agent evaluates the relevant business records and recommends actions.

The manager could then review the findings in Slack and approve the appropriate follow-up work.

This isn’t just a hypothetical direction. Salesforce has already described how Agentforce Sales and Slack can work together to identify deal risks and help employees take action.

The appeal is obvious. Employees spend less time switching applications and more time working with information they can actually use.

But I wouldn’t measure the success of that implementation by how many questions the manager asks Slackbot.

I’d want to know whether the pipeline reviews became more useful, whether opportunity records improved, and whether follow-up actions were completed more consistently.

That’s a much better measure of whether the technology is doing its job.

“If It’s Not on Salesforce, It Didn’t Happen”

There’s a saying many of us in the Salesforce ecosystem know well: “If it’s not on Salesforce, it didn’t happen.”

I’ve heard it countless times, and I understand why organizations reinforce it.

Salesforce is supposed to give teams a reliable view of their customers, opportunities, activities, and pipeline. That only works when people consistently record the information that matters.

But I also know that not everyone is going to update Salesforce every single time.

A customer conversation may occur and the sales representative may forget to log it. Someone might have an important update in Slack and not add it to the opportunity record. A decision may be made at a meeting, with the pertinent details remaining in someone’s notes.

That happens.

We can keep reminding employees to update their records, but we also have to recognize that data entry is a source of friction.

I see this as an interesting opportunity for Slackbot.

If employees are already talking about customers, projects and opportunities in Slack, can AI help connect those conversations to Salesforce?

Instead of expecting someone to remember every detail and manually update a record afterward, imagine an assistant that could help identify relevant information, prepare a suggested update, and ask the employee to confirm it before anything changes in the CRM.

That could be genuinely useful.

Of course, not every Slack conversation should become a Salesforce record. People speculate, share incomplete information, and change their minds. Organizations still need clear rules about what gets recorded and who approves it.

But there’s a difference between maintaining data governance and making data entry unnecessarily difficult.

Data becomes powerful when it’s recorded, accessible, and reliable.

And perhaps this is where Slackbot could help organizations, including yours and mine, improve how information makes its way into Salesforce.

I don’t see that as replacing the CRM. I see it as potentially removing some of the friction that prevents people from using it consistently.

After all, the goal was never to make employees spend more time updating Salesforce.

The goal was to make sure the business had the information it needed to make better decisions.

This is why I keep coming back to the importance of having a single source of truth in Salesforce.

Don’t Build an AI Agent for a Process You Haven’t Defined

One concern I have with the current enthusiasm around AI agents is how quickly organizations can move from identifying a problem to selecting a product.

A team says it spends too much time preparing reports. Someone suggests an AI agent.

Another team struggles with customer handoffs. The proposed answer is another agent.

Before long, the organization has several new automations, but the underlying problems may still be there.

Take customer handoffs.

If teams don’t define what information to transfer, assign ownership of the next step, or establish when a handoff is complete, AI workflows will struggle to deliver consistent results.

It might make the process faster. It might also make an inconsistent process happen more frequently.

Before deciding whether Slackbot or Agentforce should handle a workflow, I’d ask:

• What is the actual problem we’re trying to solve?
• Where does the information needed to complete the work come from?
• Which system owns the final business record?
• What decisions can AI make, and which ones require human approval?
• How will we know whether the workflow improved anything?

Those questions may sound basic, but they’re the difference between deploying AI because it’s available and deploying it because there’s a business case.

There’s also no reason to assume that every automation needs an AI agent.

If a task follows a consistent set of rules, an existing Salesforce Flow or another conventional automation may be easier to maintain.

AI delivers greater value when teams need to interpret context, manage variations, or make decisions that go beyond a simple set of predefined conditions.

AI Adoption Is a Change Management Challenge, Not a Finish Line

This is another area where I think organizations need to reconsider their approach.

When companies introduce a new AI solution, they often shift their attention toward adoption metrics. They start measuring how many employees use it, how many requests they generate, and how many AI agents they deploy.

I understand why companies track those numbers. They want to know whether their investment is being used.

But usage doesn’t tell the whole story.

Management may expect the employee to use an AI assistant on a daily basis. That doesn’t mean, necessarily, that the assistant is helping them work faster.

Others may use AI only a few times a week for specific tasks, but still save a considerable amount of time.

“How much value is each employee getting?

That’s why I view adoption as part of change management, not the end goal of success.

Employees need to know what the technology is supposed to improve. They need training, realistic expectations, and opportunities to give feedback when something isn’t working.

More importantly, the tool needs to fit into the way they work.

This is one reason why Slackbot interests me. If people already spend a lot of their day collaborating in Slack, then it might take less behavioral change to introduce useful AI capabilities there than to get them to learn yet another application.

But don’t mistake convenience for success.

I’d still like to know whether the implementation improved data quality, reduced unnecessary admin work, or helped employees complete important tasks more consistently.

Adoption tells us whether people are using the technology. Business outcomes tell us whether it was worth adopting.

What About Governance?

The AI assistant can operate across multiple systems, which complicates governance.

Think about a request that starts in Slack, pulls customer data from Salesforce, refers to a document sitting somewhere else, and recommends updating an opportunity record.

Who can view the information? Who decides if the action is allowed? What system? What if the recommendation is incorrect?

These are not questions that should be left until after deployment.

Both Slackbot and Agentforce have permission and security controls, but organizations still need to understand how those controls work together in their specific configuration.

An employee being able to ask for information doesn’t automatically mean an AI agent should be allowed to modify it.

I would also want clear ownership of the workflow.

If someone initiates a request in Slack that incorrectly updates a customer record, who takes responsibility for resolving the issue? Does it fall on the Slack administrator, Salesforce administrator, RevOps team, or the person managing the AI implementation?

Organizations should establish clear ownership before launching these workflows.

As Salesforce continues to introduce new Slackbot and Agentforce capabilities, teams should also verify feature availability, licensing requirements, and user permissions instead of assuming they can immediately access every newly announced feature.

Where I’d Start If I Were Planning This for a Revenue Team

I wouldn’t begin with a large deployment involving every department.

I’d choose one process that employees already find frustrating and where the organization can measure whether things improve.

Weekly pipeline reviews would be a reasonable candidate.

The work already has the Salesforce records, team discussions, reports and follow up actions. It also has a regular schedule, which makes it easier to assess.

I would begin by writing the current review process.

Teams often lack the information they need to move work forward. They should identify which questions come up repeatedly, what information employees struggle to find, and what actions they need to take after each meeting.

Then I would evaluate what can be improved through Slackbot, what needs Agentforce or other Salesforce automation, and what should be human decisions.

Slackbot may be able to facilitate the preparation of a summary of recent account discussions. Salesforce reporting would give the most current opportunity values and stage changes. A configured agent could assess particular risk conditions and suggest follow-up actions.

The manager would still decide which issues deserve attention.

After a few review cycles, I’d compare the results against the original process.

If preparation time decreased but the data was still inaccurate, I wouldn’t call that a successful implementation.

If the meetings became more focused, records were updated consistently, and follow-up work was completed more reliably, then there would be a stronger case for expanding the approach.

That gives the organization something more useful than an AI adoption metric. It gives it evidence that the workflow is improving.

Slackbot or Agentforce? Maybe We’ve Been Asking the Wrong Question.

A few years ago, companies were rushing to figure out how to introduce AI into their operations. Today, we’re starting to see what happens when AI becomes part of the tools employees already use.

I think that’s a much more interesting development.

I’m still skeptical of organizations deploying AI simply because they can. I’ve seen enough technology projects to know that buying a tool and getting people to use it are two very different things.

But I also recognize the opportunity here.

If Slackbot can help employees capture information they would otherwise forget to record, and Agentforce can help organizations act on that information through properly designed business processes, there’s a compelling case for using them together.

That doesn’t mean every workflow needs AI, or that every organization should rush to implement both products.

It means we should be looking for the points where technology removes friction without sacrificing accountability.

For me, that’s a better starting point than counting how many AI agents an organization has deployed.

The measure of a successful AI implementation shouldn’t be how much AI people use. It should be how much better the business operates because of it.

Make Your Salesforce AI Strategy Work for Your Business

AI can make Salesforce and Slack more useful, but the real value comes from connecting those capabilities to the way your organization operates.

Revenue Ops LLC helps businesses evaluate their Salesforce environments, improve data quality, and design processes that can support more effective automation.

Whether you’re considering Agentforce, exploring Slackbot, or trying to understand how both fit into your technology stack, we can help you identify where to begin.