Slack unveils Agents for Slack, letting paid users access Salesforce AI agents, third-party agents from Asana, Cohere, Adobe, and others, and users' own agents
Workplace messaging app Slack wants to make it easier to build and connect AI agents for clients.
Context & Ripple Effects
Slack had already moved from generative-AI assistance toward workplace automation: the SlackGPT plan introduced summarization and EinsteinGPT integration, while Slack AI later added thread summaries, recaps, and in-product questions for enterprise users.
This launch extends Salesforce's agent push immediately after Agentforce debuted for customer-facing business tasks. Rather than limiting agents to Salesforce-built tools, Slack is positioning its paid workspace as an access layer for partners' and customers' agents.
First-order effects
- Paid Slack customers can invoke Salesforce, partner, and self-built agents from the collaboration environment where work discussions occur.
- Salesforce and partners including Asana, Cohere, and Adobe gain a distribution point inside Slack, while customers can expose internally built agents to their teams.
Second-order effects
- Agent providers now have an incentive to make their tools callable from Slack, increasing pressure on competing workplace platforms to offer comparable agent integrations and developer paths.
- For buyers, agent selection can become more tied to workflow fit and Slack integration than to a single vendor's model or application suite.
Third-order effects
- If adoption holds, collaboration software may evolve from a destination for messages into an agentic work surface that routes tasks across multiple enterprise systems.
- The open-agent approach creates a governance challenge: organizations will need to manage which third-party and internally built agents can access workplace context and act on it.
The trend: This is one step in the shift from embedded AI features to interoperable agents operating within the daily work interface.