1. Headline & intro
Slack is no longer pretending to be just a chat app. With Salesforce’s latest update, Slack is being pushed into something far more ambitious: an AI-driven command center for your entire workday. That sounds productive – and potentially intrusive.
In this piece, we’ll look at what Salesforce is actually shipping, why this move is less about fancy features and more about strategic control of enterprise workflows, how it positions Slack against Microsoft Teams and Google, and what European companies should be asking before they switch this on.
2. The news in brief
According to TechCrunch, Salesforce has unveiled around 30 new AI features for Slack at an event in San Francisco. The centerpiece is a major upgrade to Slackbot, turning it from a simple assistant into a more autonomous AI agent.
New capabilities include “reusable AI skills” – pre-defined tasks that Slackbot can perform repeatedly across different contexts – plus the ability for companies to define their own skills. Slackbot can now act as a Model Context Protocol (MCP) client, connecting to external services and to Salesforce’s Agentforce platform, routing tasks and questions to other agents or apps.
The bot can also transcribe and summarize meetings, generate recaps with action items, and even monitor user activity across desktop tools to suggest follow-ups. Salesforce says the features will roll out over the coming months and positions them as part of its broader AI transformation, with Slack reportedly now used by around a million businesses.
3. Why this matters
This update isn’t about chat quality-of-life tweaks. It’s Salesforce declaring that the digital "place where work happens" should also be where decisions are made, tasks are delegated and follow-ups are enforced – by AI.
Reusable AI skills are essentially macros on steroids: small workflows that can be triggered in natural language, powered by company context. "Create a budget" or "prepare Q3 pipeline review" becomes a composite task: search conversations, read connected apps, draft a plan, invite the right people, schedule the meeting. Once teams adopt this pattern, they’re not just chatting in Slack; they’re encoding core processes into it. That’s incredibly sticky from a vendor-lock-in perspective.
The winners:
- Salesforce, which turns Slack from a collaboration add-on into the front door for Agentforce and its broader AI stack.
- Busy knowledge workers and managers, who get quicker access to summaries, follow-ups and routine workflows.
- Ops and RevOps teams, who can standardise processes as skills instead of endless wiki pages nobody reads.
The potential losers:
- Point-solution SaaS tools that only do meeting notes, email summaries or basic workflow automation – Slack can now bundle much of that.
- Employees if AI monitoring crosses into surveillance and performance scoring, or if automated actions create new failure modes that nobody fully understands.
In short, Salesforce is trying to move Slack from "communication layer" to "operating system for work" – and that changes the power dynamics in the enterprise stack.
4. The bigger picture
Slack’s AI-heavy makeover sits inside a wider land grab: everyone wants to own the layer where employees live all day. Microsoft is embedding Copilot into Teams, Outlook and Office documents; Google is pushing AI into Workspace and Chat. OpenAI and others are turning chat interfaces into agent hubs that talk to your tools.
Salesforce doesn’t own the operating system, the browser or the office suite. It does own CRM and, via Slack, a strong foothold in daily communication. Turning Slack into an AI agent coordinator is Salesforce’s answer to Microsoft’s deep integration advantage. If Teams is the default for many Office customers, Salesforce needs Slack to be the smarter, more flexible option – the place where multi-vendor systems still work together.
The MCP angle matters here. By letting Slackbot act as an MCP client, Salesforce is aligning with a broader trend toward open-ish standards for AI agents talking to different tools. In practice, that means Slackbot can orchestrate not just Salesforce products but any service wired into that protocol. If Salesforce executes well, Slack becomes the UI for a whole fleet of agents rather than a single closed assistant.
Historically, Slack grew on the back of its app ecosystem – bots, integrations and slash commands. The new AI skills system is that ecosystem’s next generation: instead of just posting notifications, apps can expose capabilities that an agent composes into workflows. Think of it as the bot marketplace evolving into a skill marketplace.
This points to where the industry is heading: fewer manual clicks between SaaS tabs, more AI-mediated workflows triggered in natural language. The battle is no longer “Chat vs Email,” it’s “Which AI sits in the middle of all your tools?”
5. The European / regional angle
For European organisations, the question isn’t just "Can Slack’s AI make us more productive?" It’s also "At what compliance and cultural cost?"
Desktop-level monitoring – pulling in "your deals, conversations, calendar and habits" – runs straight into Europe’s strict privacy expectations. Under GDPR, employers must have a clear legal basis and respect principles like data minimisation and purpose limitation. Works councils in Germany or France have historically pushed back hard on tools that can be used for performance surveillance. If Slack’s new AI starts feeling like a digital line manager, expect friction.
Then there’s the AI governance layer. The forthcoming EU AI Act will require more transparency and risk management around high-impact AI systems, especially if they influence hiring, promotion or performance evaluation. Even if Slack’s AI is positioned as a “productivity assistant,” the actual use in organisations will matter. If managers start relying on AI-generated metrics and summaries to judge employees, European regulators will take an interest.
Strategically, this also reinforces Europe’s dependency on US hyperscalers for core digital infrastructure. Yes, Salesforce offers EU data residency and compliance certifications, but the AI models, training pipelines and product decisions sit in San Francisco. That creates room for European alternatives – from Matrix/Element and Rocket.Chat to smaller sovereign-collaboration players – to differentiate on privacy, transparency and local control, even if they can’t match Salesforce’s feature count.
European CIOs will need to treat Slack’s new AI not just as a tool rollout, but as a cross‑functional project involving legal, HR, works councils and security from day one.
6. Looking ahead
Over the next 12–24 months, expect Slack’s AI push to unfold along a few predictable lines.
First, skills will move from demo to governance problem. Today, reusable AI skills sound like handy shortcuts. Tomorrow, they’ll represent critical business logic: how budgets are drafted, how deals are followed up, how incidents are escalated. Enterprises will need catalogues, approval flows, versioning and audit trails for these skills. Whoever controls that governance layer controls a lot of power.
Second, the commercial model will change. Bundled AI is fun for launch, but GPU bills are unforgiving. Look for Salesforce to introduce or expand AI add-ons, usage tiers, or per-seat pricing tied to advanced Slackbot capabilities. That will force hard comparisons with Microsoft 365 Copilot, which many companies are already piloting.
Third, integration depth will decide winners. If Slack can genuinely orchestrate workflows across non-Salesforce systems – Atlassian, ServiceNow, GitHub, internal tools – it strengthens its case in mixed environments where Teams is not loved but Office is entrenched. If, in practice, the best experience only exists when you’re “all-in on Salesforce,” many CIOs will see this as classic suite lock-in.
Watch for three signals:
- How granular and transparent the monitoring and data controls are for end users and admins.
- Whether Salesforce publishes clear documentation on model use, data retention and training.
- How quickly third‑party developers adopt AI skills as a first‑class integration concept.
The opportunity is enormous: a genuinely helpful, context‑aware assistant in the place you already work. The risk is an opaque automation layer that nobody fully governs until something breaks – legally, operationally or culturally.
7. The bottom line
Salesforce is making a bold, and somewhat aggressive, bid to turn Slack into the AI brain of the enterprise. If implemented with strong guardrails, it could finally deliver on the promise of less "work about work" and more actual work. Without those guardrails, it risks becoming another opaque surveillance and automation layer.
Before switching these features on, organisations – especially in Europe – should ask: who controls our new AI co‑worker, and whose interests does it really serve?
