Turning conversation into knowledge: how Slack builds human-agent teams
02:00 · August 19, 2026 · Claude Blog
Summary
Slack integrates Claude into its platform to support human-agent teams that treat workplace conversation as a growing source of institutional knowledge. Chief Product Officer Jaime Delanghe, who joined the company in 2017 to advance search and machine learning, argues that agents become effective only when they can draw on the same open context that human colleagues already use. Public-by-default channels supply this shared history, allowing agents to surface relevant decisions and avoid forcing people to repeat information that already exists in searchable records.
The daily workflow follows a repeating cycle of handoffs. Agents draft summaries, prepare meeting briefs, monitor external developments, and rewrite documents, then pass the output to a person for review and direction. The person returns the work with decisions or adjustments, and the agents continue with the next step. Delanghe describes her own Monday routine as a series of such agent-generated briefings that she evaluates before handing tasks back, illustrating how specialized agents function like teammates with defined responsibilities rather than generic chatbots.
Adoption spreads most readily when teams observe concrete examples in public channels. At both Slack and Salesforce, employees have shared working setups, debugging approaches, and workflow patterns in open spaces, allowing practices developed in one function to influence others. Measurement focuses on outcomes rather than activity metrics such as message volume or token counts, because higher activity can simply indicate that information remains hard to locate. Organizations are advised to begin with small groups in shared channels equipped with the same resources, then allow successful patterns to propagate rather than attempting to accelerate existing processes without redesigning them.
Why it matters
This article provides actionable organizational strategies for product teams looking to integrate AI agents into their daily workflows. While it lacks specific Dutch market data or deep technical code, the best practices for AI adoption, context sharing, and productivity measurement are highly applicable to Dutch SMEs and enterprise product builders.
