Who's building proactive agents? A May 2026 landscape

Six months ago, nobody was really shipping proactive agents. Now every major AI lab either has one in market or has one in internal testing. The convergence is something else: OpenAI, Google, Anthropic, Meta, Perplexity, and several startups all building agents that act without being asked.

This page maps the landscape through the three-primitives framework: does each product have a clock (scheduled execution), a listener (real-time change detection), and an inbox (multi-channel delivery)? That framework turns out to be a useful lens for separating marketing from architecture.

Last updated: May 13, 2026.

The landscape scored against three primitives.

The scorecard

Product	Company	Clock	Listener	Inbox	Stage
ChatGPT Pulse	OpenAI	✓	✗	✗	Shipped (Pro)
Orbit	Anthropic	✓	✓	Partial	Announced
Remy	Google	✓	✓	✓	Internal dogfood
Hatch	Meta	✓	✓	Partial	Internal testing
Computer / Comet	Perplexity	✓	✓	✓	Shipped
Managerbot	Block (Square)	✓	✓	✓	Open beta
Cody	CodeWords.ai	✓	Partial	Partial	Beta
Playbooks	Writer	✓	✓	✓	Shipped
Proactive AI Agents	Tonkean	✓	✓	✓	Shipped
AI Workers	MindStudio	✓	✓	✓	Shipped
Agency	Browser Use	✓	✓	✓	Shipped
Agent Relay	Agent Relay	✓	✓	✓	Shipping

A pattern is visible: nearly everyone has the clock. The listener and the inbox are where products diverge, and where the real difficulty lives.

Horizontal assistants vs. vertical agents vs. infrastructure.

The big labs

ChatGPT Pulse (OpenAI) launched in September 2025 as part of ChatGPT Pro. It processes chat history, Gmail, and Google Calendar overnight, then delivers 5 to 10 personalized morning cards. The personalization is genuinely impressive. The architecture is clock-only: no real-time change detection, no delivery outside the ChatGPT app.

Orbit (Anthropic) was unveiled at the Code with Claude conference in May 2026. It generates proactive briefings from connected tools (Gmail, Slack, GitHub, Calendar, Drive, Figma) on a timezone-aware schedule. "Orbit apps" let users pin specific insight views. It has the listener, which puts it ahead of Pulse architecturally, though the inbox is still limited to the Claude interface.

Remy (Google) is the most ambitious on paper. A Gemini-powered "24/7 personal agent" that monitors signals across Gmail, Calendar, Docs, Drive, Keep, Tasks, GitHub, WhatsApp, Spotify, and Photos. It anticipates needs and executes multi-step workflows proactively. Currently in internal dogfood (employee-only Gemini app on phones), expected to surface publicly around Google I/O. If the integration depth is real, Remy has all three primitives working within the Google ecosystem.

Hatch (Meta) is a general-purpose consumer agent trained in practice environments that replicate DoorDash, Etsy, and Reddit. It learns preferences, persists memory across conversations, and decides when to act without prompting. A separate Instagram shopping agent is also in development. Internal testing targets end of June 2026. The proactive design is there, but the delivery channels are still Meta-surface-only.

The shipping products

Perplexity has moved furthest, fastest. Perplexity Computer orchestrates 19 models for long-running autonomous tasks. Comet handles browser-based agent work. Push notifications go to email and mobile. Samsung Galaxy integration embeds Perplexity into Notes, Clock, Gallery, Reminder, and Calendar. All three primitives are present and actively used.

Block's Managerbot is the strongest vertical proof point. Built into Square POS, it monitors seller business data continuously, identifies problems, and proposes solutions. It handles shift scheduling (analyzes forecasts, generates optimized schedules), inventory management, and marketing (drafts win-back campaigns targeting the best customer segments). Available to 1M+ US sellers in open beta, with GA expected June 2026. This is proactive AI delivering measurable business value, not just interesting cards.

Writer's Playbooks are event-driven enterprise workflows: listen for triggers (emails arriving, sales calls completing, files landing), then execute multi-step automations. All three primitives, focused on enterprise teams. The least consumer-visible product on this list, and potentially the most revenue-generating.

Tonkean's Proactive AI Agents operate across 250+ enterprise systems as autonomous digital workers. Their agents run in three modes: time-based (scheduled), signal-based (event-driven change detection), and delegation-based (human-initiated). Delivery goes through Slack, Teams, email, and directly into enterprise platforms. The positioning is explicitly proactive, with agents that monitor continuously and anticipate renewals, risks, or anomalies before teams have to ask. Tonkean focuses on enterprise process orchestration, particularly procurement and operations, which gives it the same bounded-problem advantage as Managerbot and Writer.

The startups

CodeWords.ai raised a $9M seed led by Visionaries Club, with angels from the CEOs of Miro, ElevenLabs, Personio, Zalando, and Supercell. Their agent Cody takes natural-language instructions, asks clarifying questions, writes Python logic, and deploys automations on their infrastructure. 3,000+ integrations, runs 24/7. The positioning is explicitly proactive: "The best operators don't wait to be asked."

The target audience is non-technical operators, which is a different market segment than developer-focused tools. The open question is depth: 3,000+ integrations could mean deep event subscriptions or thin polling wrappers. At beta stage, there's no public data on reliability or error recovery for the 24/7 agents.

MindStudio is a no-code platform for building and deploying AI agents, with access to 200+ models and 1,000+ integrations. Agents can run on custom schedules (daily, weekly, or interval-based), trigger from webhooks and inbound emails, and deliver results via email, Slack, and SMS. The platform sits at the builder layer rather than the application layer: you construct the proactive behavior yourself rather than getting it out of the box. MindStudio also publishes consistently good content on agent design patterns and workflows, which makes their blog worth following for anyone building in this space.

Agency (Browser Use) flips the interaction model: instead of you prompting the AI, the AI prompts you. Agency watches your connected tools, identifies useful work, drafts the action, and sends it to you for one-click approval. It runs 24/7 through Telegram with thousands of integrations including Gmail, Slack, DataDog, and Linear. It works with Claude Code and Codex subscriptions, which means the agent and the app share a machine and the agent can edit and improve itself based on your decisions. All three primitives are present: scheduled monitoring, real-time change detection across integrations, and delivery through Telegram. The self-improvement loop is the interesting differentiator here.

What the convergence tells us

Every product on this list arrived at roughly the same architecture independently. Scheduled execution, change detection, multi-channel delivery. The three-primitives framework wasn't so much a prediction as a description of what proactive agents just need to work.

The differences are in coverage and depth:

• Coverage: How many data sources can the agent watch? Google has a natural advantage with Remy because it already owns the data. Everyone else is integrating third-party APIs one at a time.
• Depth: Is the listener doing real-time event streaming or periodic polling? The user experience is completely different. A four-hour-old alert is not the same as a real-time notification.
• Delivery: Can results go where the action is (Slack, email, tickets), or are they trapped in the product's own UI? Most horizontal assistants are still trapped.

I think the products that nail all three at depth are the ones people will actually reorganize their workflows around. Everyone else just ships a clock and calls it done, and those end up feeling like another tab to check.

What to watch next

Google I/O (late May 2026) will likely show Remy publicly for the first time. If Google ships deep integration across its own ecosystem, the bar for everyone else goes up immediately.

Meta Hatch targeting end of June means two billion Instagram users could have a proactive agent within months. The scale implications are hard to overstate, even if the initial capabilities are narrow.

Block Managerbot GA in June will produce the first real usage data from a vertical proactive agent at scale. If Square sellers measurably benefit, expect every vertical SaaS company to start building their own version.

Anyway, this is moving fast. I'll keep this page updated as products ship and things change.