Slack has revamped its Slackbot, which will now function as a business AI assistant integrated into many of the business processes users deal with daily. This is the next step for Slackbot, but certainly not the last.
The development of AI continues at a rapid pace this year. This year, we will see more and more business AI assistants that directly support you in your daily processes. You may be thinking: I already have Copilot, ChatGPT, or Claude, right? But the keyword for the new AI assistants is integration.
The new generation of business AI assistants
The idea behind the new generation of business AI assistants is that they integrate with a wide range of applications, enabling them to take on more work. You can think of Slackbot as your personal assistant with exactly the same rights as yo. The Slackbot can see all messages in your DMs and the channels you are a member of in Slack. It also has access to the same data you do. If Google Calendar or Outlook is linked to Slack at the organizational level, Slackbot also has access to your calendar and that of your colleagues. Slack has built a number of native integrations for Slackbot, including Google Calendar/Outlook and Salesforce.
According to Slack, the new Slackbot is also built differently. This is especially true if your organization uses Slack extensively for work, projects, and conversations with colleagues. By providing all that data as context to Slackbot, the bot knows everything about your work, what you are doing, what you still need to do, and what you may have forgotten. This allows you to work faster and more efficiently.
What does that mean in practice?
Before your vacation, you had a meeting on budget rounds and investments, and you made decisions with several colleagues. However, after your vacation, you no longer remember exactly what was decided. With one question to Slackbot, you can find out exactly what was decided and which questions may still be open within a few seconds. Previously, you had to search through the conversation history manually.
You are part of a project team with five colleagues and you need to organize a meeting next week that everyone must attend. You can use Slackbot to set a date. Slackbot has access to all calendars and can suggest several possible dates within seconds.
Slackbot can also be used to analyz s of documents, prepare conversations with colleagues, and draft reports for managers on the progress of a project: with a simple question, Slackbot can extract all the necessary information and context from Slack.
This is just the beginning; the focus still needs to be on integrations and MCP
Slackbot is really just showing a glimpse of what is truly possible. Especially if you are a Salesforce customer and use several of their solutions. Slackbot then also has access to CRM, Sales, and Service data that the Slack user also has access to. So you can also ask questions about customers or have Slackbot prepare a conversation with a customer based on recent orders and service requests.
During a call with Salesforce at, they informed us that Slackbot will support MCP, the Model Context Protocol, in the near future. This means that Slackbot will be able to communicate with other MCP servers and agents, access much more data, and perform actions via MCP. If you use a third-party CRM or ERP solution with an MCP server, it will soon be easier to integrate. We will have to wait a little longer for MCP to become available. The same applies to actions that Slackbot can perform. For example, Slackbot can currently read calendars, but cannot schedule appointments yet.
Where is Mulesoft?
Salesforce also has MuleSoft in its portfolio, an iPaaS platform that could quickly expose many third-party actions and data to Slackbot, but currently doesn’t play a role. Even though Slack says it will offer MCP support. Ignoring Mulesoft would be a big mistake. Not all applications will support MCP in the short term. That’s where Mulesoft can make a difference. We see other SaaS platforms adding tools and data very quickly using Pipedream, for example, a much smaller competitor to MuleSoft.
Finally, Slackbot is currently focused entirely on internal data, but we believe that with the right guardrails, a Perplexity integration could also be valuable. This would also make external data available that could be very useful.
It is, of course, only a matter of time before we see more from Slackbot. In April of this year, Salesforce will host its developer conference, TrailheadDX. New announcements for Slack and Slackbot are likely to be made there.
Small steps forward
What we do find positive is that Slack has chosen to improve Slackbot in small steps. What we mean by that is that they don’t say, “We have developed Slackbot as a personal assistant that can help you with everything.” Instead, they clearly communicate how it has been trained, what Slackbot can and cannot do, and what they are working towards. This makes it clear to users what is and is not possible. The actions that Slackbot can perform work well, which is extremely important for user confidence. You can also simply ask Slackbot itself, “What can you do for me?”
By comparison, we also see tech giants making AI assistants available that supposedly can do everything and integrate everywhere. This is then followed by extensive advertising, and organizations are more or less forced to adopt it. Users start working with it, but the results are disappointing time and time again. In many cases, AI ends up costing more time than it saves. These assistants will probably improve, but the question is whether users will have the patience to keep trying something over and over again if the results are disappointing.
Slack has a head start because of active users
Many tech companies are going to try to provide users with a personal business assistant. Salesforce is doing this with Slack, Google with Gemini, Microsoft with Copilot, SAP with Joule, and Workday and ServiceNow are following suit. It is not entirely clear whether ChatGPT and Claude will also get involved in this, but it would not be illogical. There are still countless startups with ambitions.
Of course, Slack has the advantage of already having users. Organizations that work with Slack already have employees who spend hours in the application every day. By adding Slackbot to this, it will be used automatically. However, Slack must ensure that it continues to make progress every few months.
In our opinion, the lack of access to external data and actions should not be delayed too long. At the moment, you can develop your own Slack application and bot to integrate with your own CRM or ERP application, for example, but this works completely alongside Slackbot and the two bots cannot work together.
Hopefully, the MCP and MuleSoft integrations will be available soon.
Slackbot available in many languages and from Business+ subscriptions
In my opinion, the biggest drawback of Slack AI to date is its limited support for transcription languages. Slack only supports a handful of languages for this. If you organize a Huddle in Dutch, for example, you won’t get a decent summary. We were a little worried that this would also apply to Slackbot, but fortunately, it speaks almost all languages. It can also analyze conversations in all kinds of languages in DMs and channels. Furthermore, Slack has decided not to charge extra for the new Slackbot, but it is only available with Business+ subscriptions. Pro subscribers have to make do with a very simple (dumb) version.
Read also: Slack claims that conversation data is a gold mine for AI agents