Anthropic has launched Claude Tag, an AI agent that works inside Slack for Team and Enterprise customers. The tool allows teams to tag @Claude in a channel and assign work without leaving Slack, representing a significant evolution from earlier AI chat interfaces. According to Reuters, Claude Tag is currently in beta for Claude Enterprise and Claude Team customers on Slack, with Anthropic planning to expand the capability to other platforms in the coming weeks. The tool can remember relevant information from the channels it's in, work under its own organisational identity, and follow up on its own tasks, making it feel less like a search box and more like a junior teammate sitting quietly in the room. 65% of the code Anthropic's product team ships is now written by an AI that lives in their Slack, demonstrating the tool's deep integration into enterprise workflows.
The latest Anthropic report analyzing 9,700 Claude user interactions reveals significant productivity gains across multiple dimensions. Respondents reported productivity gains in speed, scope and quality of work at 86%, 82% and 69% respectively, with around 27% of respondents feeling gains through cost savings on services they would otherwise have to purchase. The majority of people also reported learning more with AI (68%) and feeling like AI has made their skills more valuable (57%). In conversations for tasks tied to higher-wage occupations, Claude augments more than replaces human effort, producing higher-value outputs with the person collaborating throughout. The report notes that time of day, day of week, and even specific days of the year shape how people use Claude, with news requests peaking around 6am and recipe requests spiking to 2.3x their average at 6pm.
According to Anthropic's latest Economic Index Report, over 35% of Claude users believe AI could perform most of their work within the next year. The report examines AI's impact on the labour market and economy, focusing on how AI is being woven into the cadence of how people work and live. This edition looks at how the world comes to AI, how people put AI to work, and how they perceive its impact on work. The launch of Claude Tag demonstrates how AI is moving from "assistant" to "participant" in workplace operations, with the tool representing the third redesign of how we use LLMs - progressing from websites visited, to apps downloaded, to standing teammates with organizational tools and memory.
Based on survey analysis with 9,700 Claude users, the report found that people who delegate more to AI feel more optimistic about their futures. Most respondents expect significant near-term AI impacts to their jobs, but those who delegate the most tasks to AI are the most optimistic about job security, new opportunities, and pay. The report addresses common concerns about delegation, noting that "a commonly voiced concern about delegation is that handing entire tasks to AI means offloading thinking, with gains in output coming at the cost of learning and skill atrophy." However, the report dismisses this concern, stating that heavier delegators reported learning at the same rate as everyone else. This marks the first time an AI major has asked users directly 'how they use AI and what they feel about it' according to the report.
For South African businesses, the launch of Claude Tag raises significant questions about data access, governance, and POPIA compliance. A Slack workspace can contain client names, HR issues, sales numbers, invoices, passwords, and confidential product plans that AI tools must handle with appropriate safeguards. Companies need to establish clear rules before switching on AI tools, ensuring they process personal information lawfully and apply appropriate data handling safeguards. The tool includes spend controls so AI usage doesn't quietly run wild, and admin controls allow owners to decide which channels, tools, and repositories Claude can reach, providing some level of governance oversight. Claude Tag represents the cleanest version of AI agent ownership compared to vendor-hosted alternatives, where users rent agents and risk context pooling inside vendor walls versus self-hosted solutions that keep secrets and roadmaps on their side of the firewall.
