Runs unattended: a census of 149 listings

The vendor states its digital workers execute complex tasks without supervision, and describes Alice prospecting, handling replies and booking meetings around the clock with no human step named.

Checked 2026-09-23

The desktop layer is positioned as an autonomous agent workspace that runs overnight negotiation tasks, but the vendor states it may not place purchase orders autonomously and that sending a drafted quote can be routed through sales review.

Checked 2026-09-27

Vendor states the agent resolves the large majority of inquiries end to end across voice, messaging and email without a human; handoffs to human agents are configured for the rest.

Checked 2026-10-11

An unattended pair runs without permission prompts and auto-resumes across quota windows, but the vendor describes a human steering the agents rather than full hands-off completion.

Checked 2026-10-10

Autonomy is sold as a bounded engineering control with stop conditions and escalation, and the firm says it only raises autonomy after supervised evidence, so unattended completion is deliberately limited rather than the default.

Checked 2026-10-11

The runtime package runs an unattended observe, think, act loop until it reaches a final answer or the configured step ceiling, with no human turn required in between.

Checked 2026-10-11

Runs execute as background jobs and on schedules without a person present; a run only pauses when it hits a tool you marked as needing approval.

Checked 2026-10-11

It resolves whole conversations without a human once a Scenario is approved, but only on intents your team deployed and only when confidence clears the threshold; everything else escalates.

Checked 2026-09-23

Connected agents execute actions without a per-call human approval step, but review is after the fact in Audit and the autonomy belongs to the calling agent rather than to Loadout.

Checked 2026-10-03

It can be driven headlessly with a single message and blanket approval so one task completes without a human, but the default interactive mode is pair programming with prompts.

Checked 2026-09-30

Event Studio lets agents be triggered by system events and resolve issues without a person starting the task, and the orchestration engine plans and executes multi-step fulfilment itself.

Checked 2026-09-24

Autonomy is a setting: the default prompts before writes and commands, while always-proceed or a headless one-shot run finishes a task with nobody watching.

Checked 2026-09-10

An agent can run the whole loop unattended for free actions, but every credit-spending step and every campaign launch is gated on explicit approval, and destructive tools are annotated so clients ask first.

Checked 2026-09-26

Ava runs prospecting, sending, replies and booking without a human in the loop by default, and approvals are an optional setting you can switch on per campaign rather than a stage you graduate out of.

Checked 2026-09-10

Voice AI Agents are designed to handle an entire inbound call end to end, answering questions and completing tasks with no human agent involved.

Checked 2026-09-24

Non-interactive experts run end to end on triggers; human checkpoints are configurable, not mandatory.

Checked 2026-09-02

Microsoft states agents can act autonomously, and the Magentic-One orchestrator plans, delegates and self-reflects until a task completes without a human turn.

Checked 2026-09-19

On the phone it runs the whole job unattended - qualifies, checks the calendar and books - but on Yelp the AI only drafts a reply for staff to review, and text conversations stop after five or ten AI replies.

Checked 2026-09-11

Completing a task without a human closing the loop is the vendor's own dividing line between a chatbot and the agents they build, stated as a capability claim rather than a roadmap item.

Checked 2026-09-11

Agents run whole phone calls unattended at volume; a human only re-enters when a rail escalates or transfers.

Checked 2026-09-11

One prompt produces a whole build unattended and up to twenty prompts can be queued, but every run still starts from a human message in chat.

Checked 2026-09-11

Autonomous Nodes run an unattended reasoning loop that decides what to do next and which tools to call; handing off to a human is a configured exit, not the default.

Checked 2026-09-02

Agents run unattended: bulk outbound calling from an uploaded CSV, plus scheduled and background agents on the plan, with escalation to a human live-chat operator when needed.

Checked 2026-09-13

Managed Agents complete a prompted web task from one API call, with no human in the loop.

Checked 2026-09-02

A deployed agent runs a whole call unattended, handling turn-taking, interruptions and tool calls; it transfers to a human only when a configured condition matches.

Checked 2026-09-02

With no turn or budget cap the loop runs to completion without a human in it.

Checked 2026-09-02

Auto mode has a classifier approve tool calls instead of a person; routines and background sessions run unattended, including in the cloud.

Checked 2026-09-02

Auto-approval and headless mode let a task run start to finish with nobody watching, including scheduled cron runs and CI jobs; the vendor warns to use a clean branch.

Checked 2026-09-12

Reviews run unattended on every PR, but the coding agent stops at a proposed patch by default and a person must accept it before anything reaches the repository.

Checked 2026-09-15

Approval prompts can be switched off entirely, including a no-sandbox full-access mode.

Checked 2026-09-02

The comment, public reply and first DM run with nobody watching, but the AI layer only starts once the recipient answers the hook and stops when your handoff rule fires.

Checked 2026-09-28

The Concierge agent is described as closing issues end to end — status lookups, account updates, order changes — and handing off with full context only when it cannot resolve.

Checked 2026-09-12

AI Agents are sold on resolving conversations end to end without a human, with containment as the headline metric; complex or high-stakes moments are handed to supervisors instead.

Checked 2026-09-13

The vendor positions crews as autonomous workers that interact with enterprise applications unattended, with hierarchical delegation and cron scheduling available on both plans.

Checked 2026-09-13

Tasks complete without a human in the loop; /goal keeps the agent working towards an objective, and cloud agents run with the laptop closed.

Checked 2026-09-02

The agent handles inbound customer requests end-to-end across channels without a human, though changes the agent proposes to itself are staged for human approval first.

Checked 2026-09-14

The API runs the whole listen-think-speak loop live, handling interruptions and calling functions to take action without a human in the conversation.

Checked 2026-09-02

Automations and schedules start sessions from events, so Devin works the task with nobody tagging it.

Checked 2026-09-02

A task can be handed to a cloud Devin session that runs unattended on its own VM, while the local agent's default permission mode prompts before actions until Allow rules are set.

Checked 2026-09-03

Agents run in their own sandbox, executing commands and files to take on open-ended work; workflows also start unattended from schedule, webhook or integration triggers.

Checked 2026-09-02

Agents run unattended on schedules and event triggers, but a root task started from the web app or extension pauses for human confirmation once it reaches 600 credits of AI spend.

Checked 2026-10-01

Agents run whole calls and chats unattended, including scheduled outbound campaigns, with explicit system tools to end a call or hand off to a human number.

Checked 2026-09-02

Once a session starts the agent plans and executes end to end without supervision (you can close the tab and return to a finished output), though it may ask clarifying questions before it begins.

Checked 2026-09-14

Your agent can run a whole flow unattended over the API, but account sign-ins are a deliberate human step and the agent never gets your credentials.

Checked 2026-10-09

A Series generates and posts episodes on a schedule with no human in the loop, though the vendor also lets you approve, edit or skip each video before it goes out.

Checked 2026-10-03

Headless droid exec runs one-shot tasks to completion; Missions orchestrate long multi-feature work.

Checked 2026-09-02

Fin resolves whole conversations end to end without a teammate, including multi-step Procedures that call external systems; it escalates instead of guessing when its confidence thresholds are not met.

Checked 2026-09-18

The /agent endpoint takes a prompt with no URLs and searches and navigates on its own until it returns data, though it is an early-access preview.

Checked 2026-09-24

Automations run saved instructions on a schedule with no human present and cannot ask a clarifying question mid-run, but connector actions that send or change data default to needing approval, and the vendor frames output as work a lawyer still reviews.

Checked 2026-09-14

Opt-in: yolo mode in VS Code (trusted workspaces only) and Auto-approve changes in JetBrains remove approvals. The default is to approve each plan and tool call.

Checked 2026-09-02

Super Agent is described as running multi-step work to completion unattended, and Claw adds scheduled tasks plus a Heartbeat that acts without you starting a conversation.

Checked 2026-09-15

The cloud agent completes assigned tasks unattended in a GitHub Actions environment, capped at 59 minutes per session; Free has no cloud agent.

Checked 2026-09-02

Autonomous agents plan their own approach from plain-language instructions, and time-based triggers run them unattended.

Checked 2026-09-02

Research pages, event summaries and classifications are composed automatically with no human in the loop, but a user still chooses the market or query to analyse.

Checked 2026-10-05

Agents can be triggered by events, queues or a schedule and run as background processes with no person in the loop, which the vendor documents as ambient agents.

Checked 2026-09-15

Autonomous mode is the shipped default and needs no human in the loop; goose run takes an instruction file, allows 1000 turns without user input, and recipes can be scheduled by cron.

Checked 2026-09-16

A run plans subtopics, searches, validates sources and writes the report with no human step, and the human-feedback gate is opt-in.

Checked 2026-09-16

Plain-English triggers let agents monitor apps or the web and act with no human in the loop; scheduled and webhook triggers run unattended.

Checked 2026-09-30

Agents run whole calls unattended: greeting, understanding, checking a live calendar, booking, reminding, and rescheduling. Outbound campaigns dial, retry, and classify a list without a person in the loop.

Checked 2026-09-10

Policy decides allow or block automatically on covered events, but ambiguous actions are deliberately paused for a human approval.

Checked 2026-09-29

Triggers can send an AI answer to the customer with no agent involved, and a resolution is only counted when no human handoff was needed; low confidence hands back to a person.

Checked 2026-09-24

The agent sources, ranks, fills employer forms and follows up around the clock, but the first email batch needs explicit human approval and the sign-up page says nothing sends unapproved.

Checked 2026-09-14

The agent runs a multi-step loop on its own and a strategy can auto-approve tools outright, but the shipped default is an approval gate on anything destructive, so a session can sit waiting for you rather than completing unattended.

Checked 2026-09-17

Scheduled runs re-execute an analysis and deliver results unattended, and tool calls fire without approval, but the vendor also documents Julius pausing to ask clarifying questions mid-task.

Checked 2026-09-17

It completes multi-step tasks unattended, and a brave mode removes the approval prompts entirely; headless CI runs and GitHub Action triggers have no human in the loop at all.

Checked 2026-09-18

The assistant completes an inbound call end to end without a person on the line: it greets, qualifies, then transfers, offers a calendar slot or promises a callback, and delivers a ticket afterwards.

Checked 2026-09-26

Cloud Agents run long tasks unattended in isolated containers and open PRs.

Checked 2026-09-02

Engine scans traces and drafts fixes on a schedule with nobody watching, and deployed agents run as background or cron-scheduled jobs rather than only in a chat session.

Checked 2026-09-18

Agents built in Langflow use the LLM as a reasoning engine to choose and run tools without a human choosing each step, and the docs describe this as autonomous task execution.

Checked 2026-09-25

Built for unattended long-running work: checkpointed durable execution lets a run survive failure and resume, and human-in-the-loop checks are described as something you add rather than a default.

Checked 2026-09-02

Discovery, enrichment, scoring and drafting run end to end without a human, but the send is a hard stop: the vendor states there is no auto-send switch and launching is a human action in the app.

Checked 2026-09-18

The vendor states agents run calls and campaigns on their own and book in-home estimates end to end without a person, with escalation to a human only when the call needs one.

Checked 2026-09-09

Headless CLI runs, scheduled prompts and subagents execute without a human turn; unrestricted mode auto-allows tools.

Checked 2026-09-02

Routines run on a schedule or trigger with nobody present, and Lindy pulls data and drafts on its own, but its docs state that actions which write or send stop for a human approver first.

Checked 2026-09-19

The reading and writing are done by the agent with no human staff behind it, but every engagement is bracketed by human email: the buyer sets the question and approves the quote before work starts.

Checked 2026-10-10

Agents run as server-side workers dispatched into a live session, handling calls end to end and handing off to a human only when the flow transfers.

Checked 2026-09-02

Parse/extract/classify jobs run unattended via API, SDK or MCP; multi-step workflows are explicitly built to include human review.

Checked 2026-09-02

The vendor positions its agents as deliberately semi-autonomous: they plan and act, but sensitive steps stop at a human approval gate rather than completing unattended.

Checked 2026-09-09

Runs a task through to a finished deliverable without step-by-step supervision, and paid plans allow 20 concurrent and 20 scheduled tasks running unattended.

Checked 2026-09-19

Agents iterate their own tool-use loop until a final answer or stopping condition; durable agents persist run state so a loop can outlive the request that started it.

Checked 2026-09-02

The platform is sold on unattended resolution, claiming up to 93% of incoming queries closed without a human, with escalation triggered by configurable confidence and risk signals rather than by default.

Checked 2026-09-25

The agent-mode provider ships an explicitly autonomous execute mode that works a plan to completion, and the Harness keeps an agent progressing through long multi-step tasks without a person each turn.

Checked 2026-09-19

Agents can be triggered on a schedule, by email or by webhook and run their whole workflow without anyone present, which is the vendor's stated automation model.

Checked 2026-09-20

AI Agents reply to customers around the clock without a person present, but they only act once assigned through an automation workflow and the vendor tells you to configure a fallback and human handoff.

Checked 2026-10-03

Build mode runs a multi-step loop unattended, but Plan mode explicitly withholds any file write until the human approves and switches modes.

Checked 2026-10-09

Ambient agents run in the background on webhook or schedule triggers, monitoring and acting without anyone prompting them first.

Checked 2026-09-20

A headless run takes one prompt to completion with approvals disabled, and session goals keep the runtime queueing follow-up work until the objective is met.

Checked 2026-09-20

Agents run a reasoning loop until they produce a result and fire from schedules or channels with nobody watching; approval steps are something you add.

Checked 2026-09-02

Send mode and Autopilot dispatch applications unsupervised within a daily ceiling; Ask mode instead pauses for your approval, so autonomy is a per-account setting.

Checked 2026-09-29

Agent mode acts until the job is done without a human in the loop, and the vendor explicitly answers that babysitting is not required.

Checked 2026-10-08

The agents they build act on events without a human starting each step, but only inside confidence thresholds, escalating to a person when confidence drops or the case is novel.

Checked 2026-09-26

An agent runs whole inbound or outbound calls unattended, and workflows fire calls, messages and CRM writes from a trigger with nobody watching.

Checked 2026-09-09

The runner drives its own loop across model turns, tool calls and handoffs until the task finishes, so a run can complete with no human intervention unless you attach approvals.

Checked 2026-09-20

A session takes its own conversational turns and executes remote MCP tools itself, but your application must create the follow-up response after each tool call and owns function-tool execution.

Checked 2026-09-02

Scheduled tasks, heartbeats, background tasks, hooks and standing orders let the agent act on its own between conversations, not only when a human types a message.

Checked 2026-09-21

An --auto flag approves everything not explicitly denied, but the shipped default asks for approval on external directories and repeat loops.

Checked 2026-09-21

Event-triggered runs complete the task with nobody in the loop, and headless CLI mode forcibly approves every action.

Checked 2026-09-21

Computer takes recurring tasks and executes them unattended in the background; runs are described as lasting hours or months.

Checked 2026-09-02

Standard agents only act when a user writes to them; unattended scheduled runs arrive with Agentic Runtimes, which the docs mark as beta and gated behind a call with the vendor.

Checked 2026-09-09

A described task runs on its own schedule and returns a finished result, though sending, submission, signature and sensitive changes can be kept as human checkpoints.

Checked 2026-10-03

The voice agent conducts inbound and outbound calls unattended, handing off to a human only when its rules say to.

Checked 2026-09-02

Agents run preparation work on their own but are deliberately built to stop: anything client-facing or uncertain is queued for a named reviewer, so no task completes end-to-end without a person.

Checked 2026-10-06

Coder's default instructions push autonomous investigation through to completion, and the GitHub Agentic Workflows integration runs the same agent headless on a runner with nobody at a prompt.

Checked 2026-09-21

The vendor positions Piper as working inbound leads without a human in the loop, around the clock, across the website and the email inbox rather than drafting for a rep to send.

Checked 2026-09-22

Agent-to-tool and agent-to-agent connections can be set to Auto Run so actions execute with no human in the loop, with an optional cap on automatic executions.

Checked 2026-09-22

Background tasks run unattended in isolated copies, but nothing reaches your main version until you review and apply each result.

Checked 2026-09-22

Agents take and place whole calls without a person on the line, including scheduled outbound campaigns with pacing, retries and voicemail handling; a human can listen in and take over by choice, not by requirement.

Checked 2026-09-23

Every send and post waits for approval by default; the trust ladder can let a kind of work auto-ship after enough clean approvals, but money spend and live-conversation replies never do.

Checked 2026-10-11

The agent takes the whole call end to end without a human listening in, and the vendor's own analytics measure the share it closed alone; unresolved cases escalate rather than stall.

Checked 2026-09-09

Agents fire on schedules, deal events, transcript completion and webhooks, and auto-prospecting can enrol good-fit contacts without a manual pass; Rox's own engineering docs still say critical actions are queued for human review.

Checked 2026-09-03

The receptionist handles routine questions and collects booking requests on its own, but the vendor frames the end of every flow as a handoff to human staff rather than completed work.

Checked 2026-09-28

Autonomy is a per-agent setting rather than a fixed trait; agents escalate to a human when an issue exceeds their scope.

Checked 2026-09-03

Scheduled work executes in the cloud without the agent running, but the docs still direct confirmation of message content and approval controls exist for sensitive actions.

Checked 2026-09-25

The vendor's FAQ states the default mode is full autopilot with no human in the loop; a human may optionally assist, approve or take control, but nothing requires it.

Checked 2026-09-25

Unattended operation is real but deliberately bounded: agents arrive switched off, and even in the highest mode new content, redirects, canonicals, noindex and pruning always wait for a person.

Checked 2026-10-03

Agents resolve end-to-end customer requests without a human in the loop, including email replies and multi-step actions in systems of record.

Checked 2026-09-26

Schedule triggers run a deployed workflow on a timer with no input and no operator present; webhook triggers fire on external events.

Checked 2026-09-03

Flow runs a whole call unattended (greeting, transcription, speech, transfer, hangup), but every decision comes from your application, so autonomy depends on the logic you supply.

Checked 2026-09-26

Scheduled audits run unattended on a cron expression with webhook alerts, but every result is a finding a person must read and act on; nothing is applied automatically.

Checked 2026-10-03

The agent completes whole inbound calls unattended: answering, qualifying, booking into the calendar and notifying the owner afterwards rather than during.

Checked 2026-09-26

Background agents fire on schedules, CRM events or webhooks and complete the job unattended; in chat it also runs multi-step work end to end without approving each step.

Checked 2026-09-15

Trigger nodes start workflows on external events or on a schedule with no operator present, and human review is an optional node rather than a requirement.

Checked 2026-09-09

Agents answer and place calls unattended, escalating to a human only on triggers you configure.

Checked 2026-09-03

An agent completes a task unattended and posts its report, but it must stop at In review and cannot approve its own result, so a person or a separate reviewer agent closes the loop.

Checked 2026-10-03

Agent-OS agents run on their own schedule without anyone starting them, delivering outputs by email or connector; pilot runs are supervised first and the client approves outcomes rather than each step.

Checked 2026-09-12

Agents run whole calls and write to connected systems without a person on each interaction; escalation rules and human handoff conditions are set during agent setup rather than removed.

Checked 2026-09-16

Once configured, the assistant answers, speaks, routes and invokes tools through a full call without a human in the loop; a human enters only if the agent uses the transfer or handoff tool.

Checked 2026-09-09

Given a goal it navigates, adapts and continues until the goal is complete.

Checked 2026-09-03

Autonomous is one of the two agent types you pick at creation, human escalation is marked optional in the build flow, and a deployed agent runs as an unattended Orchestrator job; UiPath itself steers high-stakes regulated work back to deterministic automation.

Checked 2026-09-03

The deep search plans its own query sequence, follows citations and authors and decides itself when to stop, with no human step between submission and result.

Checked 2026-09-26

Agents research continuously without a human present, but signal-triggered automations and automated plays that act end to end are Business-tier only.

Checked 2026-09-27

Agents take whole calls end to end and escalate to a human only when the configured path says so.

Checked 2026-09-03

Agents run tasks and scheduled work on their own, but the vendor states that important or sensitive actions can still stop and wait for a person to approve them.

Checked 2026-09-27

Heartbeats, schedules, watchers and subagents run unattended in the background, but at the default risk tolerance anything that changes state still stops for your approval.

Checked 2026-09-27

Vera handles whole inbound calls with no human present, resolving routine questions and taking messages, and outbound AI calls run a goal-directed conversation end to end before reporting back.

Checked 2026-09-22

Agents answer and place calls unattended and hand off to a human only when they cannot handle the caller.

Checked 2026-09-03

The assistant completes an inbound call end to end — greeting, FAQ answers, detail capture, booking request — and only pulls a human in when its configured rules say to, which is the vendor's stated design.

Checked 2026-10-03

The GTM Brain layer decides and executes without waiting for a human to act on a dashboard.

Checked 2026-09-03

Cloud agents run unattended from schedules, integrations, CI or the API; locally, Run until completion auto-approves every command until the task ends.

Checked 2026-09-03

The stated default is an unattended run: an agent starts research by tool call, the engine keeps working to the budget, and results return as JSON.

Checked 2026-09-27

Reads, writes and ordinary project commands run automatically, but a human must sign in to the web AI first and approve anything sensitive or destructive.

Checked 2026-09-21

The agent works a task through to completion on its own, but permission-gated tool calls default to asking first, so unattended running requires setting tool_permissions.default to allow.

Checked 2026-09-29

Agents complete tasks unattended on a schedule or from a repository event, load context, edit, run tests and publish results without a person approving each step.

Checked 2026-09-09

Vendor states the agents resolve inquiries and execute multi-step workflows without a live agent; escalation to humans is a configured fallback.

Checked 2026-09-03

Confirmation is opt-in rather than default, and the headless modes deliberately run tools without asking so pipelines complete unattended, bounded only by a step cap.

Checked 2026-09-28

The agents they build are described as completing multi-step workflows end to end, with human-in-the-loop escalation reserved for edge cases rather than every step.

Checked 2026-09-25