Capabilities / Autonomy & background work

Personal AI agents that work in the background

As of 9 Oct 2026, 30 of the 30 profiled personal AI agents have documented or observed records for scheduled, triggered and long-running work that continues without you. 2 of them have independent observations; the rest rest on vendor documentation or partial evidence. 0 profiled products have no record yet, which means unknown, not unsupported.

Availability is not task success: a documented capability says nothing about how reliably the agent completes the task. For measured performance see the leaderboards.

Independently observed (2)

Instinct

Instinct (Spear Street Technology) · Universal Personal Execution Agent

Third party records Independently observed
During the observation in September, duplicate car rentals and abnormal expenses were proactively discovered; the 7 o'clock briefing on the working day of September 11 was on time for two consecutive mornings.source ↗
Official trigger and management interface specifications Not disclosed
The listed sources were reviewed, but sufficient public documentation was not found.source ↗

Ollie

Confabulation Corporation · Home personal assistant

Background monitoring Documented
Keep an eye on your email, calendar and to-do list, and get text message reminders when important; the official website has daily briefings.source ↗
Regular measurement Independently observed
The 9/17 record weather reminder on the list will be sent within minutes on the hour for three consecutive days at 9 o'clock on working days.source ↗

Documented by the vendor (17)

Meta Muse

Meta · Persistent personal agent

Background and proactivity Documented
Continue after the app is closed; proactively identify emails to handle, make suggestions, and follow up based on ongoing goals.source ↗
Schedules and events Limited or conditional
The documentation is clear about cron/alerts/monitoring; some connectors will proactively update agents for calendar changes.source ↗

OpenAI dots

OpenAI · Persistent personal agent

Proactive vs authorized work Documented
Active research has read-only tool restrictions; background tasks after authorization can perform actions according to normal approval rules, and the two must be displayed separately.source ↗
Scheduled event followthrough Limited or conditional
It supports fixed schedules, available service event triggers, and can also be automatically paused/waked up according to task status to follow up; only connecting services will not automatically create monitoring.source ↗

Lindy(Personal Lindy + Teammate)

Lindy · Individual/Team Agent

Timing and event proactive Documented
Routines are composed of triggers, prompts and destinations; there are daily/weekly/monthly or email sending and receiving, Slack messages, pre-meeting/post-meeting events, and Smart Filter filtering; chat, text…source ↗
Built-in follow-up Documented
Built-in daily briefing, email classification/draft, urgent email reminder, unanswered email follow-up draft, meeting minutes/preparation/arrangement; can be turned off individually.source ↗

Town

Town.com · Work-based personal agents and team automation

Timing and events Documented
Supports scheduling, email/calendar events, and webhooks; Starter only supports schedules and can be triggered once a day at most, Pro and above have all types of triggers; Free weekly newsletter only.source ↗

Pally

Pally Technologies · Multi-channel universal personal execution agent

Initiative and trigger Documented
Morning briefs, important message reminders, scheduled tasks, and event-triggered custom automations, such as notifications when a specified sender receives a letter.source ↗

Asaply

Asaply Inc. · Local consumption/errand transaction type personal agent

Nudge routine Documented
The user gives a weekly meal ordering routine, sends a confirmation when arriving, and orders the meal after replying yes; it is not the default that no one confirms and the payment is automatically deducted.source ↗

Caddy

Caddy Technologies Inc. · Message-Based Universal Personal/Home Execution Agent

Routines and events Documented
Temporary/cyclic reminder; new users default to the local 8 o'clock morning briefing, which will be sent only if it is useful. Important changes in Gmail/Calendar/Slack/Linear can trigger check-in.source ↗
Controllable initiative Documented
Active messages can be stopped while retaining individually created reminders and routines; only selected connection sources are checked.source ↗

OpenClaw

OpenClaw Foundation · Self-Hosted Personal Agent

Execution environment Documented
User-managed gateway and tooling environment; available in remote or local models.source ↗
Scheduled background Documented
Persistent cron can wake up the agent and deliver to chat or webhook; there is also a heartbeat/event entry.source ↗

Hermes Agent

Nous Research · Self-Hosted Personal Agent

Execution environment Documented
Native/VPS or isolated backend; documentation lists Docker, SSH, Singularity, Modal, Daytona, Vercel Sandbox.source ↗
Scheduled background Documented
Supports one-time/repeated cron, paused editing and result delivery; background review can maintain memory/skills.source ↗

AutoGLM (hosted application)

Zhipu / Z.ai · Hosted cloud-device agent

Execution environment Documented
Executed on a mobile phone/computer in the cloud data center, it does not occupy the local mobile phone execution interface.source ↗
Background Documented
iOS 2.0.12 (2025-11-23) has declared the background execution of Life Assistant.source ↗

Doubao Phone Assistant (consumer version)

ByteDance · Device-integrated personal agent

Execution environment Documented
Native system/application operation, combined with client-side understanding and cloud model; supports AI key, voice, and lock screen controlled entry.source ↗
Background Limited or conditional
The official website lists multi-task queuing, insertion, shortcut commands and automatic commands; mobile operation is still marked as Beta.source ↗

Qwen App (task execution / work assistant)

Alibaba · Consumer ecosystem personal agent

Execution environment Documented
Consumption services are integrated with the cloud ecosystem; the work assistant can connect to the PC to process local files/commands after scanning the QR code for verification and authorization.source ↗
Scheduled background Historical or unconfirmed
In August, it was reported that scheduled tasks and office assistants were added; work can be delivered at the agreed time.source ↗

Coze 3.0 / Coze Agent (formerly Coze Space)

ByteDance · Hosted personal work agent

Execution environment Documented
The native agent is resident in the cloud; it can be used on cloud devices or connected to local devices for execution.source ↗
Scheduled background Documented
Agent scheduled tasks replace the old Agent Plan; background long tasks can continue while other sessions are working.source ↗
Constraints Documented
Cloud devices do not support some code scanning/face recognition logins, overseas sites, and any system modifications; automation may trigger third-party risk control.source ↗

HONOR YOYO Agent

Honor · Device-integrated personal agent

Execution environment Documented
Native system/App operation, cloud inference when necessary; invoked and granted scope of access by user.source ↗
Contextual suggestions Documented
YOYO suggestions can provide service suggestions for pickup, payment codes, flights and other scenarios.source ↗

MiniMax Agent/MiniMax Code Work Mode

MiniMax · Hybrid general-purpose work agent

Execution environment Documented
Web uses cloud environment; desktop work can use local files/browsers; context can be reviewed and then transferred to the cloud.source ↗
Scheduled background Documented
The desktop can plan daily/weekly reports and other recurring tasks.source ↗

QwenWork

Alibaba / DingTalk · Hybrid personal and team work agent

Execution environment Documented
The desktop mainly operates local files/browsers; the web handles cloud tasks and full-stack pages.source ↗
Cloud scheduled Documented
Web scheduled tasks run in the cloud. They can still be executed after logging out/turning off the computer, and there are task records.source ↗
Desktop scheduled Documented
Desktop scheduled tasks are scheduled by the machine according to the system time zone, and local files, connectors, and skills can be used.source ↗

WorkBuddy (Tencent)

Tencent · Desktop personal work agent

Execution environment Documented
Mainly use the local working directory and tools of the desktop, and the mobile phone/IM can be distributed to the desktop.source ↗
Scheduled background Documented
Automatically configure prompt words, directories, time and models/skills/connectors, and the results can be synchronized with mini programs/enterprise WeChat.source ↗

Partly disclosed or tested (3)

szn

Nomadic Futures · Universal Life Execution Agent

Follow up Partly disclosed / partly tested
The official website states that it can continue to follow up/review tomorrow/proactively inform; the specific timing and event triggering configuration have not been publicly disclosed and fully explained.source ↗
Long-term monitoring and actual measurement Partly disclosed / partly tested
9/19 Observation proactively reminds unpaid voyages, cancellation deadlines and check-ins, and acknowledges that the voyage has passed when making corrections; the routine is not rated.source ↗

Shuffle

DNFT · Creation/entertainment companion assistant, including website execution

Backend capabilities Partly disclosed / partly tested
The policy describes scheduled nudges/reminders, execution progress and active follow-up, which can be reduced or turned off; the public information does not give the event trigger editor details.source ↗

Tomo

Mapo Labs · Personal goals/life companionship and execution assistant

Reminders and Initiatives Partly disclosed / partly tested
The official website claims that it can proactively send messages and reminders; HealthKit data can drive goal progress and proactive suggestions; any webhook/timing workflow specifications are not given.source ↗

Limited or conditional (8)

Grok Bot

xAI (via Cursor) · Individual/Team Agent

Routines events Limited or conditional
Supports background timing and supported events/independent webhooks; event connections are different from ordinary plug-ins; a single Bot can have up to 50 routines, a minimum of 5 minutes, and retains the latest 20…source ↗
Workflow learning Limited or conditional
Shared skills can be saved from successful work; Teach a task generates skills to be verified with a screen demonstration of up to 10 minutes, which will be gradually opened.source ↗
Unattended approval Limited or conditional
Background approval initiated by non-current user chats expires in about 10 minutes and will not be executed; long-term inactivity may ask and suspend routines.source ↗

Claude (including Cowork execution)

Anthropic · General-purpose work agent

Backstage and planning Limited or conditional
All paid plans support scheduled tasks; new cloud tasks can be run even after shutdown, and can be paused/resumed/manually run/viewed past results. Tasks that require local resources are still subject to device online…source ↗
Any event triggered needs to be verified Unknown
This time to confirm the scheduled task, personal Claude's native trigger matrix for any connection service event has not been confirmed.source ↗

Manus 2.0 / Manus Studio

Manus · General-purpose work agent

Timing and events Limited or conditional
Automations include timing, event triggering and advanced beta workflows; support email, calendar, Slack/Notion and GitHub/Shopify/Webhook, etc. Specific events/conditions depend on the connector. Some are near…source ↗
Background execution and maintainability Limited or conditional
Automation can view time/status/results, edit/pause/test; repeated failures may be automatically paused. You can choose to continue the original task or a new task (depending on portal availability), and…source ↗

Gemini Spark

Google · Persistent personal agent

Background/timing/event Limited or conditional
Supports single and periodic, Gmail conditions, and topic event monitoring; up to 50 active schedules and 15 parallel tasks. Full quota/full concurrency will be skipped; triggering is approximate and not suitable for…source ↗

Perplexity Computer (Cloud)

Perplexity · Cloud execution agent

Timing and events Limited or conditional
The current Automations replace the old Perplexity Tasks and can be run and notified in a single time/cycle; up to 15 schedules per session. The Computer master file also lists conditions such as emails, calendars,…source ↗

Perplexity Personal Computer (local access)

Perplexity · Cloud + local execution agent

Background/event/timing Limited or conditional
The Computer backend and automation are still used; tasks related to local resources rely on the availability of connected devices. Portable provides local scheduling and persistent queues.source ↗

Microsoft Copilot Cowork for personal accounts(Preview)

Microsoft · Personal Execution Agent (Preview)

Backend/Plan Limited or conditional
You can schedule one-time or periodic tasks, confirm before activation, and then pause/edit/cancel; you can view waiting for input, running, completed and scheduled in Mytasks. Personal event trigger list not found.source ↗

Cue by Manus

Manus · Independent identity, multi-agent collaborative personal execution agent

Background timing/event Limited or conditional
Supports scheduling and supported event/condition triggering to notify or perform allowed actions; can monitor fares, reservation vacancies, and deadlines.source ↗

Other capabilities

Personal AI agents with cloud or local computers Personal AI agents that make calls and send messages Personal AI agents with app connectors and tools Personal AI agents with long-term memory Personal AI agents in chat apps, voice and other channels Personal AI agents for research and deliverables Personal AI agents that book and buy Personal AI agents for teams and parallel work Personal AI agent reliability and recovery Personal AI agent permissions, privacy and security