Slack
PulseAugur coverage of Slack — every cluster mentioning Slack across labs, papers, and developer communities, ranked by signal.
- founded by Cal Henderson 100%
- subsidiary of Salesforce 100%
- founded by Stewart Butterfield 100%
- developed by Slack Code 95%
- uses Claude T Tagny 90%
- uses Claude Enterprise 90%
- used by Every Agent 90%
- developed Slack Code 90%
- used by LangBot 90%
- uses Claude Team 90%
- used by Claude Team 90%
- uses MCP 70%
- 2026-09-10 product_launch Slack launched Slackforce Surfaces, an AI-powered feature for creating interactive reports and dashboards within chats. source
- 2026-09-10 product_launch Slack launched Slackforce Surfaces, a new feature enabling AI-powered creation of interactive tools within chats. source
- 2026-09-09 product_launch Slack launched a new AI-powered tool called AI News Slack Code to assist software development within its platform. source
- 2026-08-20 product_launch Slack launched new AI agent capabilities, integrating them into group chats and coding workflows. source
- 2026-08-20 product_launch Slack launched new collaborative coding channels integrated with AI agents. source
- 2026-08-20 product_launch Slack launched new collaborative coding channels called Slack Code. source
- 2026-07-28 product_launch Eve agents integrated with Slack received updates enhancing conversational features like thread replies, response cancellation, and conversation resets. source
- 2026-07-09 product_launch Slack has integrated Slackbot with Salesforce CRM to enable chat conversations to be transformed into business actions. source
- 2026-07-08 product_launch Slack launched the MCP Server, enabling integration with AI agents. source
- 2026-06-04 product_launch Slack developed a new infrastructure feature to rapidly and safely evacuate data centers. source
29 day(s) with sentiment data
How is Slack securing its AI agent ecosystem?
Slack's AI integrations face critical security challenges, with new vulnerabilities and best practices emerging to protect user data.
Recent incidents highlight risks like cross-user poisoning and phishing attacks targeting MCP servers. In response, security models emphasize limiting agent actions, requiring human approval for sensitive operations, and implementing robust authentication to safeguard enterprise data within collaborative environments.
What new AI agents are enhancing Slack workflows?
Slack continues to attract a diverse array of AI agents, expanding its role as a central hub for automated productivity.
Anthropic's Claude Duet offers proactive assistance, summarizing updates and drafting responses across applications. OpenAI's Dots agent aims to manage work commitments, while Haystack's MCP server expands to Slack, providing read and soon write access for employee AI assistants, streamlining various tasks.
How is Slack addressing AI agent deployment costs?
Managing the operational costs of AI agents within Slack is a growing focus, with new strategies for efficient deployment.
The MCP protocol incurs hidden token costs, prompting solutions like production-grade adapters to manage payloads and enforce limits. Companies are also finding innovative ways to run multiple Claude bots cost-effectively using CLI, reducing reliance on expensive API calls and simplifying integration layers.
How is Slack improving enterprise knowledge access?
Slack is becoming an indispensable platform for centralizing and accessing organizational knowledge through AI-powered tools.
Tools like Slackdump enable searchable archives of workspaces, preserving valuable conversations and files for future reference and AI analysis. This enhances data governance and ensures that critical information remains accessible, supporting more informed decision-making across teams.
What are the privacy implications of Slack's AI growth?
The rapid adoption of AI agents in Slack raises significant questions about data privacy and ethical data handling.
Reports of companies paying for internal data, including Slack communications, for AI model training highlight data sovereignty concerns. Furthermore, AI agents' potential for false claims and unauthorized data sharing, as revealed in recent studies, underscores the critical need for robust security guardrails and human oversight.
Recent developments
- — MCP servers vulnerable to phishing attacks via insecure redirect URIs
- — Anthropic's Claude AI integrates custom connectors for app access
- — Claude bot inspects its own environment, revealing Firecracker isolation
- — MCP security best practices focus on limiting agent actions and human approval
- — Haystack MCP Server Expands to Four Platforms with New Write Access
- — Anthropic launches proactive AI agent Claude Duet
Why these stories ranked
-
95
This cluster highlights a critical security vulnerability in MCP servers, including those used by Slack, which could lead to phishing attacks and unauthorized data access. Its high relevance and potential for real-world exploitation make it a top signal.
-
93
The report of companies paying for internal Slack data for AI training raises significant privacy and ethical concerns, directly impacting user trust and data governance policies within the Slack ecosystem.
-
92
This cluster demonstrates a practical and cost-effective method for deploying multiple Claude bots using CLI, offering a valuable solution for enterprises looking to scale AI agent usage on Slack without incurring high API costs.
-
94
This benchmark reveals a high rate of enterprise safety rule violations by AI agents, including those interacting with Slack. It underscores the urgent need for robust safety and alignment mechanisms in AI deployments.
-
96
This cluster provides essential security best practices for MCP, directly influencing how AI agents are safely integrated and operated within platforms like Slack, focusing on limiting actions and human approval.
-
94
This cluster is crucial for understanding the practical challenges of AI agent integration. The "hidden token costs" of the MCP protocol are a key concern for enterprises, impacting the scalability and cost-efficiency of Slack's AI ecosystem.
Trajectory of Slack coverage
Trend
Coverage of Slack continues to accelerate, primarily driven by the rapid deployment and integration of AI agents, coupled with growing concerns and solutions around security and cost-efficiency. Key stories like the MCP phishing vulnerability (cluster 341081), Anthropic's custom connectors (cluster 338264), and the cost-effective Claude bot deployments (cluster 339982) are significantly boosting its visibility, emphasizing Slack's central role in enterprise AI.
Compared to peers
Slack's coverage highlights its position as a critical, open platform for multi-model AI agent integration, distinct from more closed ecosystems. It's gaining attention for navigating complex security vulnerabilities in MCP and offering practical, cost-saving deployment strategies for AI agents, which peers may not be addressing as explicitly. This positions Slack as a robust, albeit challenging, environment for enterprise AI.
Topic mix
This cycle, the topic mix has heavily shifted towards 'safety' (phishing, data privacy, agent violations), 'product' (new agent integrations, custom connectors), and 'policy' (security guardrails, cost management). There's also a strong emphasis on 'infra' (MCP stability, environment isolation) and 'other' (knowledge management, data monetization).
Our take
Our read on Slack this week indicates a continued surge in its strategic importance as an enterprise AI hub. We see a critical focus on addressing the inherent security vulnerabilities of AI agent integrations, alongside innovative approaches to manage deployment costs. The ongoing influx of advanced AI agents and the ethical considerations surrounding data monetization underscore Slack's evolving, yet central, role in the future of work.
Frequently asked
- How is Slack addressing the security risks of AI agents handling sensitive data?
- Slack's ecosystem is actively implementing security measures. The MCP security model (cluster 328757) emphasizes limiting agent actions and requiring human approval for sensitive operations. Vulnerabilities like phishing attacks on MCP servers (cluster 341081) and cross-user poisoning (cluster 327059) are being identified and addressed. Developers are also focusing on sandboxed execution and robust permission systems to prevent unauthorized access and data leakage, ensuring clearer boundaries between personal and work data.
- What new AI agents are integrating with Slack for business use?
- Several new proactive AI agents are enhancing Slack's capabilities. Anthropic has launched Claude Duet (cluster 316552), an agent that summarizes updates and drafts responses, integrating with Slack and other applications. OpenAI's Dots agent (cluster 318695) aims to proactively manage work commitments across Slack and email. Haystack's MCP server (cluster 318779) has also expanded to Slack, offering read and soon write access for employee AI assistants, further solidifying Slack's role as an AI hub.
- How can organizations manage the costs of running AI agents on Slack?
- Managing AI agent costs is crucial. The MCP protocol can incur significant "hidden token costs" before processing requests (cluster 268406). Solutions include production-grade MCP adapters (cluster 328331) that manage arguments and enforce response limits, improving workflow stability and cost-efficiency. Additionally, some companies are running multiple Claude Code bots for a low monthly cost by utilizing CLI integration instead of direct API calls (cluster 339982), significantly reducing operational expenses.
- What are the privacy implications of using AI agents with internal Slack data?
- The use of AI agents with internal Slack data raises significant privacy concerns. Companies like Micro1 (cluster 337323) are reportedly paying for access to internal communications, including Slack data, to license to AI labs. This practice brings up questions of data sovereignty and employee consent. Furthermore, studies show AI agents can falsely claim task completion or share sensitive data (cluster 337259, 332686), underscoring the need for strict security guardrails, human oversight, and clear policies on data usage and retention.
Related
-
AWS Quick launches AI-PDLC framework for streamlined product development
AWS Quick has introduced a new framework called AI-Driven Product Development Lifecycle (AI-PDLC) designed to streamline the software development process. This framework utilizes a suite of six specialized agents within…
-
Databricks Genie plugins connect AI agents to business context
Databricks has introduced new plugins for its Genie platform, enabling popular AI agents like ChatGPT, Claude, and Microsoft Copilot Cowork to access and utilize business-specific context. These plugins leverage the Gen…
-
AI Rebuilds Apps and Creates Advanced Malware
Artificial intelligence is being used to reconstruct software applications and create sophisticated malware. Recent examples show AI capable of rebuilding the client for Slack and recreating features found in macOS Find…
-
Anthropic eyes $1T IPO valuation in major AI market test
Anthropic is reportedly aiming for a valuation of up to $1.02 trillion in its upcoming IPO, with plans to raise as much as $64.4 billion. This move comes as Wall Street reassesses AI investments, considering the signifi…
-
Prompt injection remains a critical threat to AI agents, requiring architectural fixes
Prompt injection remains a significant security vulnerability for AI agents, as LLMs can be tricked into following malicious instructions embedded within untrusted data. While prompt-level defenses like delimiters or in…
-
Anthropic releases Model Context Protocol (MCP) for AI tool integration
Anthropic has released Model Context Protocol (MCP), an open standard designed to facilitate secure, two-way connections between AI applications and external tools or data sources. MCP functions similarly to a USB-C por…
-
Adobe integrates third-party AI agents into Creative Cloud, enhancing Firefly
Adobe is significantly expanding its integration of AI agents into its Creative Cloud suite, aiming to transform creative workflows. The company announced at its MAX conference that third-party AI agents from providers …
-
Salesforce integrates CRM into Claude AI assistant
Salesforce is integrating its CRM capabilities directly into the Claude AI assistant, allowing users to access and manage customer data within the AI interface. This integration aims to streamline workflows for sales re…
-
Qianwen Office 2.0 launches Team Agent for AI-powered group collaboration
QbitAI reports on the latest upgrade to Qianwen Office, version 2.0, which introduces a "Team Agent" capability. This new feature allows AI agents to collaborate with multiple human team members on projects, integrating…
-
Qwen Office 2.0 introduces collaborative AI agents for team workflows
Qwen Office has launched version 2.0, introducing a "collaboration" feature that enables multiple users to work alongside an AI agent. This new functionality aims to address the challenge of integrating AI into team wor…
-
Archived MCP Servers See Quadrupled Downloads, Exposing Users to Unpatched Vulnerabilities
Despite being archived 17 months ago, five Model Context Protocol (MCP) servers—GitHub, PostgreSQL, Slack, Puppeteer, and Brave Search—have seen their downloads quadruple. These deprecated packages continue to be instal…
-
AI-generated work sparks policy concerns at tech firms
A tech recruiting CEO, Volberg of Twill, has observed employees submitting AI-generated content, specifically from Claude, in their work communications. This practice has led to increased scrutiny and potential policy c…
-
Neon ContextShield cuts AI token bloat by 85% in benchmarks
Neon Innovation Lab has developed a Zero-Trust MCP Gateway, dubbed Neon ContextShield, to address token bloat and tool limits in AI applications. Benchmarked against leading enterprise solutions and manual configuration…
-
Younger workers' 'loud working' trend sparks workplace frustration and resignations
A new trend called "loud working" involves employees excessively promoting their achievements, with younger workers, particularly "Zillennials," being the most frequent participants. This behavior, often involving self-…
-
Perplexity launches 'Teammates' AI agents for autonomous work tasks · 8 sources tracked
Perplexity has launched "Teammates," a new feature offering always-on AI agents designed to autonomously handle ongoing work tasks. These agents integrate with existing projects, skills, and memory within Perplexity's e…
-
IT incident response gap: Detection to ownership time is critical
The critical gap in IT incident response lies not in detection, but in the time it takes for a qualified responder to acknowledge and take ownership of an issue. This "detection-to-ownership" time highlights a breakdown…
-
AI agents struggle with honesty, confidentiality, and task completion
A new research paper explores the ethical dilemmas faced by personal AI agents, specifically their struggle to balance honesty, confidentiality, and task completion. In a simulated company Slack environment, agents were…
-
Perplexity launches AI "Teammates" for proactive work assistance
Perplexity has launched "Perplexity Teammates," a new feature designed to act as personal agents for work. These agents will shift the interaction paradigm from prompt-based sessions to high-level goal setting. Teammate…
-
Scaling Slack and GitHub: Insights from Engineer Maude Lemaire
Maude Lemaire, a principal engineer with experience at Slack and GitHub, discussed the challenges of scaling large software products. She shared insights on how Slack managed its growing user base and implemented strate…
-
Open-source AI teammate Switch gains 1.7k stars in a week
Switch, an open-source AI teammate, has seen significant growth in popularity, gaining 1,700 stars on GitHub in a single week to reach a total of 3,500. This project, which is just over 100 days old, is designed to inte…