SANDBASE · SCHEDULE AGENT WORK

Periodic Agent JobsDelivered on time every time.

Describe the schedule once, and SandBase will trigger each Agent run, keep execution records, and deliver results to the right place.

Schedule Agent Jobs ›View Business Cases ↘

01 · Set Run Schedule

In One SentenceSet the Agent's Run Time.

No need to understand Cron. Describe when work should occur, and SandBase will generate clear execution frequency, timezone, and next run time.

Describe run time in natural language

⌁

Schedule GeneratedReady

Agent: Market Intelligence Agent
Run Rhythm: Every morning at 9 AM
Cron: 0 9 * * 1-5
Timezone: Asia/Tokyo

About to Run

01MON · 09:0002TUE · 09:0003WED · 09:00

02 · TASK ≠ SCHEDULE

Same task,can be configured with multiple schedules.

The same Agent defines the work, inputs, and success criteria; different Schedules handle frequency, time zones, environment, and customer. Each trigger creates an independent Agent Run.

AGENT · REUSABLE WORK

Market Intelligence AgentVersion 12 · Production

JOB: Research market changes
INPUT: Market · Region · Sources
RESULT: Briefing with cited intelligence

Additional Timeline→1 AGENTOne-to-Many

SCHEDULES · WHEN / FOR WHOM

Tokyo Daily BriefingWeekdays · 09:00 · Asia/Tokyo

Enabled

London Weekly ReviewMonday · 08:30 · Europe/London

Enabled

Customer-Specific MonitoringEvery 6 hours · customer_2048

Enabled

A Run is Created Each Trigger

RUN #1082 COMPLETEDRUN #1083 QUEUEDRUN #1084 NEXT · 09:00

Results, Trace, and Costs are Recorded Separately →

03 · Business Timeline

Not running code on a scheduleBut continuously completing business work.

Choose a real scenario to see how SandBase combines data, Agent judgment, and team actions into a service chain that can run long-term and be audited.

Real-time configurationEnabled

Stripe

→

HubSpot

→

Slack

Runtime: Every weekday · 08:30 · Customer timezone
Deliverables: Renewal Risk List by Priority
Business Value: Earlier Involvement with High-Risk Customers

schedule.yaml

agent: renewal-riskschedule: weekdays 08:30timezone: customer.timezonealert: risk_score >= 0.72

04 · Manage Timeline

Define a Task Once,Schedule it with different rhythms.

Agent decides what to do; Schedule decides when, for whom, and in which environment to run. Fixed tasks can be version-controlled, and business schedules can be adjusted anytime via the Dashboard or API.

Versioned Schedule

Managed along with Agent configuration

Ideal for internal research, monitoring, and reporting tasks. Run frequency, time zones, and Agent version are all maintained in a single reproducible configuration.

Start with a fixed plan ↗

schedule.tsVERSIONED

schedule(({
  agent: "market-intelligence",
  cron: "0 9 * * 1-5",
  timezone: "Asia/Tokyo",
  version: "v12"
}));

Console Schedule

Adjust timing without redeploying

Create, enable, pause, and modify Schedule in Dashboard. Natural language generates clear Cron and next run time.

Open Schedule Console ↗

Schedule CreatedDRAFT

AGENTRenewal Risk AgentDescribe Runtime RhythmCheck Renewal Risk Every 6 Hours Generate

CRON0 */6 * * *TIMEZONEUTC

API / SDK Schedule

Dynamically create schedules for each customer

Products can be bound to different accounts with frequency, timezone, and external identifiers via API or SDK, while avoiding duplicate creation of the same plan.

View API/SDK integration ↗

create-schedule.tsSDK

await schedules.create(({
  agent: "renewal-risk",
  cron: "0 0 * * *",
  timezone: customer.timezone,
  externalId: customer.id,
  deduplicationKey: `renewal-${customer.id}`
}));

05 · Monitor each run

Schedules are responsible for triggeringEvery result is clearly visible with monitoring.

Each Schedule trigger becomes a trackable Agent Run. View execution steps in real time, receive notifications on failure, and locate issues from the same trace.

01 · LIVE STATUS

Confirm the current status of each task first

Running, queued, completed, and failed use the same status language. Teams can determine which scheduled task requires intervention without opening logs first.

Real-time
Environment Awareness
View History

Runtime MonitoringLIVE

market-intelligence()RUNNING00:42

renewal-risk()FAILED

daily-brief()COMPLETED01:18

customer-watch()QUEUED—

02 · EXECUTION TRACE

Trace every step along the Agent's actual execution path

Model calls, tool actions, retries, and outputs are aligned on the same timeline. Time consumption and bottlenecks are no longer hidden in scattered logs.

RUN #1082 · MARKET INTELLIGENCEEXECUTING

schedule.triggerexa.searchmodel.reasonnotion.create

03 · ALERTS

Deliver notifications to the right people when failures occur

Set email, Slack, or webhook alerts based on environment and event. Alerts directly link to the corresponding run, not just telling you "something went wrong."

NEW ALERTEnabled

CHANNEL#agent-operations

ALERT WHEN✓ Agent runs fail✓ Retry limit reached

Environment✓ PRODUCTION○ STAGING

04 · INVESTIGATE

Filter anomalies and continue investigating from failed steps

Filter run records by status, Agent, environment, and time. Failed reasons, inputs, and retry history are always retained in the same execution record.

Monitor Agent runs ↗

View HistoryFAILED · 3

RUNAGENTDURATIONSTATUS#1087renewal-risk00:18FAILED#1083market-watch01:42FAILED#1071daily-brief00:34FAILED

03 · Run and Improve→Next Step

The schedule triggers runs, and monitoring verifies that work is complete.

View models, tools, deliverables, costs, and anomalies per each Run.

Monitor Agent Runs →Return Agent Configuration ↗