ibl.ai's agent Tasks tab gives an agent a named prompt, a start date and time, and a cadence of once, daily, weekly or monthly, and logs every run with its status, its start and end times and the full output, on a platform where you own all the code and the data.
This is a closer look at a shipped capability, not a launch. The /iblai-vibe-agent-task skill that mounts the tab in any app built with Agentic Vibe was released in iblai/vibe v1.6.0 on July 27, 2026; the same tab runs in Agentic OS.
Can an AI agent run a task on a schedule without being prompted?
Yes. Every agent in ibl.ai has a Tasks tab, in the Runtime group of its settings beside Memory, History, Audit and Analytics. The panel says what it is for:
"Let your agent work on its own schedule. Set up recurring tasks it runs automatically — like daily summaries or reminders — without anyone having to ask."
Above the list are a search box, a Select date filter and the Schedule Task button.
Three metric cards count the agent's tasks: Total Tasks, and how many are Completed or Failed going by each task's latest run, so a schedule that has started failing is visible before anyone opens a log.
How do you schedule an AI agent task?
Click Schedule Task. The dialog has a calendar on the left — past days are disabled — and the task's settings on the right.
Task Name labels it. Task Prompt is the instruction the agent carries out on every run. Time sets when; the date and time combine into a start time that must be in the future, or the dialog will not save.
Repeat has four settings: Don't repeat, Daily, Weekly and Monthly. Each maps to a schedule on the platform: Don't repeat creates a one-off task pinned to the chosen date, which runs once and is then switched off by the scheduler.
The other three create a recurring cron schedule at the chosen time: Daily every day, Weekly on the start date's weekday, and Monthly on the same day of the month.
How do you know whether a scheduled agent run worked?
Every task in the list carries a badge with one of five states: Scheduled, Running, Completed, Failed or Disabled. The badge comes from the task's latest run together with whether it is switched on, so a one-off that ran successfully shows Completed, not Disabled.
Click a task and its run logs open in the panel beside the list, five to a page. Click a run and Log Details opens.
Log Details shows the run's status and when it was created, started and ended, followed by the agent's full output for that run.
That record is what you check when someone asks what the agent actually produced last Tuesday — and the agent's Audit tab sits one click away in the same Runtime group.
Can you add scheduled agent tasks to your own app?
Yes. The tab ships as one component, <AgentTasksTab>, from @iblai/iblai-js/web-containers/next. No props are required: it reads the organization, the agent and the user from the shared agent-settings provider, and handles listing, creating, deleting and run logs itself.
Two props are optional: labels, which overrides any string in the tab's ten label groups (header, toolbar, metrics, list, logs, states, the schedule and delete dialogs, log details and toasts), and PaginationComponent, which swaps in your own pager.
The /iblai-vibe-agent-task skill wires it up from a coding agent.
The screen-by-screen walkthrough is in Agent Settings: Tasks. To test an agent's answers before it runs unattended, see benchmarks and LLM judges for agents.
Which jobs would you hand to an agent on a schedule?
Tell us the report, summary or reminder, and we'll show it scheduled and logged on a live agent. Book a 30-minute demo or talk to the ibl.ai team — ibl.ai is family-owned and operated from New York, NY.