Slack reveals its playbook for generative AI post-redesign

Most workplace tools treat generative AI like an awkward add-on pinned to a sidebar. Slack wants to weave it directly into the bones of how teams actually get work done.

Following its newly announced interface redesign, Slack's senior director of product management, Olivia Grace, spoke with Gizmodo about where the platform is headed. The goal isn't just to add a chatbot to the chat app, but to turn communication into an operational engine.

Prototyping inside the product

Instead of pitching static slide decks, Slack tests new features right inside the live software under an internal rule called prototyping the path. If an experiment works in daily practice, it stays.

That practical approach is already evident in Workflow Builder. Users currently launch 10.5 million workflows in Slack every single week. More strikingly, 80% of the people building those automations have no technical background at all.

Recent upgrades introduced conditional logic and shareable workflow links that can be dropped into channels or direct messages. Delivery service Shipt used these automations to triple its IT response speed, while Verizon's wireline operations rely on them to coordinate field technicians on service calls.

A three-layer AI roadmap

Slack is laying out its generative AI strategy in three distinct buckets.

The first layer is native functionality inside Slack, starting with conversational summaries and message drafting. Further out, the company wants generative models to analyze channel chatter and actively suggest pre-built workflows based on what teammates are discussing.

The second layer relies on external integrations across Slack's catalog of more than 2,600 apps, letting teams bring in third-party tools like Anthropic's Claude. The third turns Slack into the primary interface for Salesforce Customer 360, pulling real-time customer data directly into team channels.

While emoji reactions keep the platform feeling like a digital watercooler, Grace emphasized that Slack operates primarily as a company command center.

Why it matters

The biggest bottleneck for workplace AI isn't capability; it is the friction of switching tabs. Because employees already spend their entire workday inside Slack, embedding AI directly into ongoing discussions eliminates the mental overhead of separate prompting.

We think Slack's smartest move is tethering generative AI to its no-code workflow engine. Summaries are handy, but having an AI watch a chaotic conversation and generate an executable workflow hands real leverage to non-technical staff.

The challenge will be maintaining Slack's trademark responsiveness without overwhelming users with automated clutter, especially as it absorbs heavier enterprise data from Salesforce.