Salesforce Just Gave Slack 30 New AI Features. Nobody Asked for Most of Them.

There is a specific type of Silicon Valley announcement that appears to be revolutionary but, for the majority of those who receive it, is more likely to cause mild confusion. Salesforce’s most recent initiative with Slack is in that range; it’s impressive in terms of engineering detail, ambitious in terms of language, and a little overwhelming in terms of volume.

Slackbot now has thirty new AI features all at once. Thirty. The majority of teams are still learning how to properly mute notifications.

To be fair, there are some truly intriguing things that Salesforce revealed last week. For example, the ability to transcribe meetings solves a genuine problem. The appeal will make sense to anyone who has listened to a forty-minute call, taken three incomplete sentences of notes, and then spent the following morning attempting to piece together what was actually decided.

Now, Slackbot listens in the background, records decisions, presents action items, and, as soon as the call concludes, drops a structured summary straight into Slack. There is nothing new to install for teams that are already using the Slack desktop application. It’s clean. In fact, that part makes sense.

However, there is the remainder of it. AI abilities that are reusable. a client layer for MCP. Slack has a built-in native CRM. A desktop agent that tracks users through all open apps on their screen, observing context changes in real time and offering to take action on anything that is visible. In a product brief, every feature makes sense.

When taken as a whole, they give the clear impression that the company has been sitting on a year’s worth of roadmap items and decided to announce them all at once in the hopes that momentum would take the place of curation.

How much of this will actually be used by employees is an honest question that no one seems to be raising loud enough. Features designed to impress procurement teams, rather than the people who work eight hours a day, have a long and somewhat depressing history in enterprise software.

Salesforce’s own statistics, which show teams saving up to twenty hours a week and $6.4 million in internal productivity value, are impressive, but they come from within the company, where adoption is probably encouraged with some zeal. There may be a big gap between a software company’s internal usage and a regional insurance company’s or a mid-sized logistics company’s actual adoption.

Whether the native CRM integration is a breakthrough or a category collision is still up for debate. The CTO and co-founder of Salesforce, Parker Harris, reportedly questioned why anyone should still need to log into Salesforce. It’s a thought-provoking statement that would likely work well in a keynote address.

However, the idea that Slack is now the front door to all of that is difficult for businesses that have invested years and significant funds in developing Salesforce workflows. Before any of this goes live, several administrators will want to have a very specific conversation about permissions, configurations, and change management procedures.

However, the Engine case study merits consideration. In just twelve days, Engine, a travel and spend management platform, developed its first AI agent on Slackbot, which ultimately resolved over half of all customer travel cases without the need for human intervention. Previously restricted to manually sampling 1% of calls, quality assurance now covers 100% of calls automatically.

These numbers show a real operational change, not marketing gimmicks. And to put it simply, context is where intelligence resides, according to Engine CEO Elia Wallen. Unlike most AI tools, Slackbot doesn’t start from scratch because it carries the entire weight of customer history and conversation data. That distinction is more significant than it may seem at first.

The fragmentation issue in enterprise AI, where each department uses its own tools and none of them communicate coherently with one another, seems to be something Salesforce is attempting to address. It is conceptually elegant to think of Slackbot as a routing layer that intercepts requests and locates the appropriate agent or application without requiring the employee to know what lives where. It is another matter entirely whether elegance can withstand contact with the organizational complexity of a real enterprise.

As this develops, it is difficult to ignore the fact that the announcement’s quietest feature—memory—is also the most likely to alter day-to-day behavior. With time, Slackbot becomes more adept at understanding each user’s preferences, shortcuts, and recurrent patterns. That isn’t a feature that makes headlines. It doesn’t fit neatly on a slide.

However, if it functions as promised, it is the kind of capability that could make everything else feel less like a product and more like something that truly comprehends the shape of your day. That’s what distinguishes a helpful tool from one that just gathers digital dust.

Salesforce is wagering that the adoption issue that has been subtly undermining enterprise AI for years will be resolved by integrating intelligence directly into the interface where work already takes place. They might be correct.

It’s also possible that adding thirty features all at once to a tool that people primarily use to share memes and gripe about Monday meetings is just asking for too much change to happen too quickly. The factories are operating. Shipping is one of the features. It’s genuinely unclear if anyone will thank them for it.