[Digital Today reporter Seulgi Son] Global CRM company Salesforce said on Wednesday it has enabled two-way integration linking Anthropic's AI assistant Claude with its collaboration tool Slack and its AI agent platform Agentforce 360.

The integration was implemented through Anthropic's open AI integration standard, the Model Context Protocol (MCP). MCP is a protocol that lets AI models access data and tools in external systems in a standardised way.

Users can search Slack conversations in Claude and use them to draft meeting agendas or planning documents, then automatically share the generated documents back to Slack channels. For example, users can handle a request such as "Summarise last week's Slack discussion by the marketing team and share it in that channel" in one go in Claude.

Salesforce will also add Agentforce 360 integration features soon. It said actions suggested by Claude will be able to be executed immediately in actual work processes such as customer management and service ticket handling in Salesforce CRM.

On security, the Anthropic model runs within Salesforce's "Trust Layer" and data is processed in customers' virtual private cloud (VPC). Salesforce stressed that the existing permission system applies as-is, making it usable in regulated industries such as finance and healthcare.

Salesforce Korea CEO Sejin Park (박세진) said, "We are focused on supporting the transition to an 'agentic enterprise' in which AI executes actual business tasks."