Automation

Microsoft officially launched Agent 365 on May 1, 2026 as a dedicated governance and security control plane for enterprise AI agents — a standalone product priced at $15 per user per month that sits above individual Copilot features. The product addresses a critical gap that has slowed enterprise AI adoption: the absence of centralized visibility into what AI agents are doing across an organization. Agent 365 provides observability dashboards, insider risk management for automated tasks, access controls for third-party agents, and audit trails that compliance teams need to satisfy regulators. It integrates with Microsoft Copilot Studio, Azure AI Foundry, and third-party agent platforms, meaning it can govern agents built anywhere in the Microsoft ecosystem — not just first-party Copilot features. Microsoft simultaneously launched the E7 "Frontier Suite" at $99 per user per month, bundling Agent 365 with E5, Microsoft 365 Copilot, and Entra Suite at a claimed 15% saving versus buying each separately. For enterprises scaling from AI experiments to production deployments, Agent 365 provides the governance infrastructure that makes deploying agents at department scale legally and operationally defensible.

OpenAI released the Privacy Filter on April 22, 2026 — an open-weight model specifically designed to detect and redact personally identifiable information (PII) from text with accuracy that exceeds previous open-source alternatives on standard PII detection benchmarks. The model ships open-weight, meaning organizations can run it on their own infrastructure without sending sensitive text to an external API — a critical requirement for healthcare, legal, financial, and HR workflows that handle regulated personal data. For enterprise automation pipelines that process customer communications, intake forms, support tickets, or employee records, Privacy Filter provides a production-ready PII scrubbing layer that can run inline before data reaches downstream AI models, reducing the risk of accidentally exposing personal data through model inputs or outputs. The open-weight release puts professional-grade PII detection within reach of any engineering team, including those operating in air-gapped or compliance-restricted environments.

OpenAI launched workspace agents in ChatGPT on April 22, 2026, giving Business, Enterprise, Education, and Teachers plan subscribers AI agents that autonomously execute multi-step workflows across connected business tools — without a human in the loop for each step. Powered by Codex and running in the cloud rather than inside a single chat session, the agents can persist across hours-long tasks such as lead outreach, pull request review, weekly reporting, and vendor risk monitoring, picking up context between sessions automatically. Slack integration shipped at launch, with Gmail, GitHub, and Google Drive connections entering preview — marking OpenAI's most direct move into enterprise workflow automation and its clearest answer to Microsoft Copilot's integrated workflow features for teams that want autonomous task execution rather than AI-assisted drafting.

Google's Agent-to-Agent (A2A) protocol reached v1.0 and entered production at 150 organizations in the week of May 5, 2026 — a milestone that signals the shift from experimental inter-agent communication to standardized infrastructure. A2A defines how AI agents from different vendors and platforms communicate with each other: how they pass tasks, share state, request capabilities, and return results. Before A2A, multi-agent systems required custom integration work for every agent-to-agent handoff, making complex automation pipelines brittle and expensive to maintain. With v1.0 stable, any agent built on an A2A-compatible platform can hand work to any other A2A-compatible agent, regardless of which vendor built it. The production deployments at Google Cloud Next 2026 demonstrated agents from Box, Workday, Salesforce, and ServiceNow operating together in automated workflows — enterprise software categories that previously required extensive custom middleware to connect. The Anthropic Model Context Protocol (MCP), which reached 97 million installs in March 2026 and was recently donated to the Linux Foundation's Agentic AI Foundation, handles agent-to-tool connections. A2A handles agent-to-agent connections. Together, these two protocols are becoming the foundational plumbing of the agentic AI era — the equivalent of HTTP and REST for the previous generation of software. For engineering teams building automation workflows, adopting A2A-compatible tooling now avoids a migration cost later as the standard solidifies.

Microsoft launched Agent 365 on May 1, 2026 — a standalone enterprise product priced at $15 per user per month that acts as a control plane for managing, deploying, and orchestrating AI agents across an organization. Unlike Copilot Wave 3, which updated the existing Microsoft 365 experience in March, Agent 365 is a separate product designed specifically for enterprises that want to run multiple AI agents simultaneously across different business workflows. The headline feature is multi-model orchestration: Agent 365 can route tasks to Claude, GPT-5.5, or Microsoft's own MAI models depending on what the task requires, switching between models automatically based on cost, speed, and capability trade-offs. For IT departments, the governance layer is the practical selling point — administrators get visibility into every agent session, can set access policies by department or user role, and receive audit logs that satisfy the documentation requirements of enterprise compliance teams. Business application teams can also use Agent 365 to build agents directly in SharePoint through a public preview powered by Claude, without writing code. Microsoft is framing Agent 365 as the management layer that sits above all the individual Copilot and third-party AI tools a company might use — a single control point rather than a fragmented collection of AI subscriptions. For companies already running Microsoft 365, the $15 per user monthly add-on price is positioned to be easier to justify than standalone agent platforms from new entrants. This is one of the most significant Microsoft product launches of 2026 for enterprise AI adoption.

Merck and Google Cloud announced a landmark partnership at Cloud Next 2026 on April 22 to embed agentic AI across Merck's entire research and commercial organization. The scale is large: Merck employs approximately 67,000 people globally, and the initial rollout targets 14,000 researchers and commercial teams in its highest-priority launch markets. The technology at the center is Gemini Enterprise. Merck's research teams will use AI agents to synthesize scientific literature, flag regulatory risks in clinical trial designs, and identify patterns across experimental data that would take months of manual analysis. Their commercial teams — the people running drug launches and managing physician relationships — will use agents to surface market intelligence, track competitor activity, and draft briefing materials. Merck's chief information and digital officer, Dave Williams, framed it as "entering an intelligent agentic ecosystem where AI works alongside our teams as we enter one of the most significant launch periods in our company's history." This is not a pilot or a proof of concept. Merck has committed to migrating core digital infrastructure to Google Cloud as part of the deal — a multi-year, multi-hundred-million-dollar technology overhaul. The pharmaceutical industry has been slower than finance or retail to adopt enterprise AI at scale, largely due to regulatory sensitivity around patient data and clinical claims. This partnership signals that the regulatory risk is now manageable, and competitive pressure to adopt AI in drug development has become impossible to ignore. Expect other major pharma companies to announce similar partnerships in Q2 and Q3 2026.

Anthropic announced Project Glasswing on April 24 — a research initiative built around Claude Mythos, its most powerful model to date, which the company has decided is too dangerous to release to the general public. Instead, Mythos Preview is being shared exclusively with a curated list of cybersecurity organizations, including JPMorganChase and select government infrastructure partners, specifically for defensive security work. The reason for the restricted release is unusual candor from an AI lab: Anthropic believes Mythos has capabilities that could be misused for offensive cyberattacks if made broadly available. Rather than wait for those capabilities to be discovered and exploited by bad actors, Anthropic is deploying them first in controlled environments to build defenses. In practice, Mythos can find vulnerabilities in critical software — the kind of flaws that take human security researchers months to uncover — in hours. JPMorganChase is using it to stress-test financial system security. Government partners are evaluating its use for protecting power grids and water supply digital systems. Anthropic says it plans to use learnings from Project Glasswing to build safety guardrails into an upcoming public Claude Opus model, after which broader access may be considered. This is a meaningful departure from the usual AI product launch playbook: instead of releasing first and asking questions later, Anthropic is explicitly acknowledging the risk and choosing a restricted rollout. For security professionals, this signals that AI-assisted vulnerability discovery is no longer theoretical — it's happening now in production environments at major financial institutions.

HubSpot announced on April 19 a radical pricing shift for its AI customer service product: instead of $500–5,000/month for a fixed license, customers now pay $0.50 per successfully resolved customer conversation and $1.00 per qualified sales lead generated by the AI agent. The premise: "You shouldn't pay for software you don't use effectively." A mid-market company handling 10,000 customer support inquiries monthly pays only for the roughly 70% the AI handles autonomously (7,000 conversations × $0.50 = $3,500/month). Worst case, they pay nothing if the AI agent is ineffective. Best case, they save $10,000+/month in agent time. Early adoption numbers: 450 customers switched in the first 48 hours, with average payback periods under 3 months. HubSpot reports a 67% deflection rate (customers resolve issues without talking to a human agent), meaning a typical customer saves 2–3 FTEs (full-time employees) worth of labor annually. The competitive shift: this model is spreading fast. Intercom, Zendesk, and Freshworks all announced outcome-based pricing trials within 6 hours of HubSpot's announcement. The broader implication: enterprise software is shifting from "buy access" to "buy results." Any AI vendor not moving to outcome-based pricing will look outdated by Q3 2026. For customers, this is the first time AI automation pricing actually aligns with business value. For vendors, margin pressure is real—they win only when customers win. This is the pricing model that breaks open AI adoption in SMB.

Anthropic's Model Context Protocol crossed 97 million installs in March—a milestone that signals MCP's transition from experimental to foundational infrastructure across the AI industry. Every major AI provider now ships MCP-compatible tooling: OpenAI has MCP support in ChatGPT Enterprise, Google integrated MCP into Vertex AI agents, Meta embedded MCP in its developer dashboard, and Microsoft standardized MCP in Copilot Pro. The Linux Foundation took Anthropic's MCP under open governance on April 18, cementing it as industry-wide shared infrastructure. What this means in practice: instead of building custom API integrations for every tool a customer wants an AI agent to use, developers now just point the agent at an MCP server. A company runs a private MCP server on their network that connects an agent to their CRM, billing system, knowledge base, and chat history—instantly, no custom code. The scaling impact is massive: a SaaS company with 50,000 customers previously needed custom integrations with each customer's tools (ERP, CRM, accounting software). With MCP, the SaaS can publish one MCP server, and 50,000 customers connect autonomously. Estimated TAM effect: agent infrastructure market goes from $5 billion (2025) to $80+ billion (2027+) because MCP eliminated the integration bottleneck. For small and mid-market tech: MCP server packages are emerging as a new business category—Zapier, Make, and Workato are all releasing MCP servers. First-mover advantage exists now; every SaaS that publishes an MCP server becomes the default integration layer for its category.

OpenAI launched GPT-5.4-Cyber on April 14—a fine-tuned version of its GPT-5.4 flagship model built specifically for defensive cybersecurity work. Unlike standard GPT-5.4, this "cyber-permissive" variant has lower refusal boundaries for legitimate security tasks including vulnerability research, binary reverse engineering, and zero-day analysis. The rollout is gated through OpenAI's expanded Trusted Access for Cyber (TAC) program, which now admits thousands of verified individual defenders and hundreds of security teams. The headline new capability: security professionals can analyze compiled software for malware and vulnerabilities without needing the original source code—a major time-saver for incident response. For context: Anthropic's Claude Mythos Preview found 27-year-old bugs in OpenBSD and chained Linux kernel vulnerabilities for privilege escalation; GPT-5.4-Cyber is OpenAI's direct response one week later. OpenAI's Codex Security has already contributed fixes to more than 3,000 critical and high-severity vulnerabilities in the wild. Access requires identity verification; higher TAC tiers unlock more permissive capabilities. The strategic rationale mirrors Anthropic's: by deploying these models exclusively to verified defenders, both companies are positioning themselves as the infrastructure layer for enterprise security—a market worth over $200 billion annually.

HubSpot released a remote Model Context Protocol (MCP) server on April 13, enabling AI coding assistants and autonomous agents to query, create, and update HubSpot CRM data using plain natural language. The integration works with any MCP-compatible AI tool—Claude, Cursor, Windsurf, GitHub Copilot—allowing developers and sales teams to access their CRM data without writing custom API code. Practical example: a developer using Claude can ask "which deals in our pipeline have been stalled for more than two weeks?" and receive a structured answer pulled directly from HubSpot, then immediately instruct the agent to update those deal statuses and log follow-up tasks. The MCP server supports reading contacts, creating deals, updating pipeline stages, pulling reports, and managing workflows. HubSpot has 220,000+ customers in 135+ countries; the MCP server multiplies the effective user base by making CRM data accessible to AI agents without manual exports or custom integrations. The timing is significant: Salesforce's AI-enhanced Slackbot uses MCP to update CRM records from conversational messages in Slack. HubSpot's MCP server gives the 220,000+ businesses on HubSpot equivalent AI-mediated CRM automation—leveling the field with enterprise Salesforce deployments at a fraction of the cost. The integration is free for existing HubSpot subscribers and requires only a few minutes of setup via the standard developer MCP documentation.

Anthropic launched Claude Managed Agents, a product that handles the hardest part of running AI agents in production—reliability, error recovery, tool orchestration, and compliance logging. Enterprises can deploy AI agents directly within Anthropic's platform without building custom infrastructure for monitoring, rate limiting, or failover handling. The product targets a recurring complaint from enterprise AI teams: building a Claude-powered agent is easy, but keeping it reliable in production at scale is a full-time engineering job. Claude Managed Agents bundles that operational complexity into a managed service. Use cases already in production: automated customer support ticket routing, code review pipelines, financial document analysis, and sales research workflows. The commercial logic is clear—once enterprise workflows run on Anthropic's managed infrastructure, switching costs increase significantly. The launch comes the same week Anthropic disclosed $30 billion in annualized revenue, up from $9 billion just four months prior.

Salesforce announced a sweeping overhaul of Slackbot, transforming it from a passive Q&A tool into an autonomous work assistant capable of executing tasks across a user's entire digital workspace. The 30 new AI features include: reusable AI "skills" that remember how you handle recurring tasks, cross-app orchestration via Model Context Protocol (MCP) enabling Slackbot to update Salesforce CRM, create Jira tickets, and schedule calendar events in one instruction, and autonomous desktop operation for handling workflows without switching apps. Use cases transforming enterprise productivity: customer escalation routing (Slackbot reads support tickets, creates tickets, pages on-call engineers automatically), pipeline management (updates CRM entries from deal notes), and meeting-to-action conversion (creates tasks from meeting transcripts). For global enterprises, MCP compatibility means Slackbot connects to any internal system without custom integration work—particularly valuable for organizations with diverse regional tech stacks.

Natural language automation is here. Tell Zapier "when a customer emails about a refund, create a support ticket and notify the team" and it builds the entire workflow.

The automation platform plans to use funding to add native AI model hosting, so workflows can run AI inference without external API calls. 500,000+ teams already use the platform.