Job scheduling for Windows servers
Outgrown Task Scheduler? You don't need Control-M.
ThinkTorch runs your scripts on your own Windows servers, on schedules and in chains. No ports to open: the agent only dials out. When a job fails at 2 AM, the email tells you why, and you rerun it from the run page.
Free forever · no card · 2 agents included
Subject [Prod] Nightly DB backup failed, exit code 1
Runs Nightly DB backup
Nightly DB backup
- Prod
- 2 m 0 s
- exit 1
Failed with exit code 1 at 02:02
Backup failed: not enough space on D: (needed 42 GB, free 3 GB).
Other recent runs
- Succeeded 07:41 · 2 m 4 s
- Succeeded 02:00 · 1 m 58 s
- Succeeded 02:00 · 2 m 3 s
What runs, and when?
Anything you can start from a command line: a script, a program, a file transfer. On a schedule in plain words, after another job finishes, when a file lands in a folder on an agent's computer, or when a webhook is called. Retries are set per job, and an optional workspace time limit stops a runaway run.
- Backups
- ETL chains
- Health checks
- Reports
- Housekeeping
| Last run | Name | Schedule | Runs after |
|---|---|---|---|
| Succeeded | Nightly DB backup | Daily 02:00 UTC | |
| Succeeded | Archive backup to NAS File transfer | Not scheduled | Nightly DB backup, on success |
| Running | Hourly sales report export | 5 * * * * | |
| Succeeded | Sync price list | Every 6h |
Flow Nightly DB backup on success Archive backup to NAS
File watch prod-app-01 · D:\incoming\prices · *.csv one run per file
Where does it run, and who can touch it?
On your own computers. Each one runs a small agent that dials out to ThinkTorch and asks for work, so there are no inbound ports to open. The job runs there, against your own files; ThinkTorch keeps the schedule, the status, the command and its output.
Environments are a wall. A Prod job is never claimed by a Staging agent, even with matching tags, and only an Admin can mint the key that puts an agent in Prod. Access is set per environment: Viewer, Operator, Editor or Admin.
Changes to jobs, runs, agents, environments, roles, keys and invites are recorded, along with attempts that were refused: who, what, where and when, in an audit log per workspace.
File transfers of a folder or a pattern need a Windows agent.
Prod
Nightly DB backup needs db
- prod-db-01 claims it
- prod-app-01
Staging
- stg-db-01 never: other environment
- stg-app-01
- Viewer
- Operator
- Editor
- Admin
| When | 09:14 |
|---|---|
| Who | sam@northwind.test |
| Action | job.updated |
| Subject | Nightly DB backup |
| Environment | Prod |
| What happened | Tried to edit job 'Nightly DB backup' without the Editor role. |
| Outcome | Refused |
What happens when it breaks?
The failure finds you, says why, and is one click from fixed.
It lights up.
Attention Failed 1Failed Nightly DB backup
You get an email.
Subject [Prod] Nightly DB backup failed, exit code 1
To ops@northwind.test
Sent to the notifications address you set, once any retries are used up.
It says why.
FaultFailed with exit code 1 at 02:02
Backup failed: not enough space on D: (needed 42 GB, free 3 GB).
02:00:00 Connecting to SQL01 as svc_backup (password ***).Search the log, follow it live, download it.
You rerun it.
RerunRerun in one click, or let auto-retry go first. An optional workspace time limit stops a runaway run, and a run whose agent dies is marked Failed instead of hanging.
- Predictable One run, one agent. Each run is claimed by exactly one agent, atomically, and a Prod job never runs outside Prod.
- Secrets Secrets are encrypted at rest and masked as *** in run logs.
Past Task Scheduler, nowhere near Control-M
| ThinkTorch | Task Scheduler | Enterprise (Control-M, AutoSys) | |
|---|---|---|---|
| Setup | Install an agent, done | Per server, task by task | Weeks, often a consultant |
| Failure alerts | An email when a run fails for good | None built in | Yes |
| Retries & timeouts | Retries per job, and a workspace time limit | Restart and time-limit settings per task | Yes |
| Job dependencies | Chain on success or failure | Not built in | Yes |
| History & logs | Every run, in one place | Per server, if history is on; no output | Yes |
| Audit trail | Changes and refused attempts | Change events per server, if history is on | Yes |
| Secrets in logs | Encrypted, masked as *** | Unmasked, wherever your script logs | Yes |
| Runs on your servers | Yes - agent is outbound-only | Yes | Yes, installed by you |
| Price | Free to start, from $19/mo | Included with Windows | $$$$ + procurement |
Simple, honest pricing
Start free, upgrade when you outgrow it. Every plan has every feature - you only pay for scale.
- 2 agents
- 1 environments
- 5 jobs
- 2 seats
- Shortest schedule: every 10 minutes
- Log history: 7 days
- 5 agents
- 2 environments
- 20 jobs
- 3 seats
- Shortest schedule: every 5 minutes
- Log history: 30 days
- 15 agents
- 5 environments
- 100 jobs
- 10 seats
- Shortest schedule: every minute
- Log history: 90 days
- 40 agents
- Unlimited environments
- Unlimited jobs
- 30 seats
- Shortest schedule: every minute
- Log history: 1 year
- Unlimited agents
- Unlimited environments
- Unlimited jobs
- Unlimited seats
- Shortest schedule: every minute
- Log history: Unlimited
Prices in USD. Every plan includes every feature - cancel anytime.
Questions, answered
Something not covered here? The Help guide explains agents, jobs, schedules, alerts and security in full.
Do I have to open any firewall ports?
No. The agent only makes outbound connections to ThinkTorch and asks for work, so there's nothing inbound to expose.
Where do my jobs actually run?
On computers you own, via the agent you install: the job runs there, against your own files. ThinkTorch stores the schedule, the status, each job's command, and the output the agent sends back, and keeps that output for your plan's log-history window after the run finishes (Free 7 days, Starter 30 days, Team 90 days, Business 1 year, Enterprise unlimited), then deletes it.
What can a job be?
Anything you can launch from a command line: a PowerShell or Python script, a batch file, or any executable, with the arguments you give it.
What happens when a job fails?
It retries first, if you set retries. If the last attempt fails too, it emails the notifications address you set and shows on the Attention page, with a link straight to the log.
Is there really a free plan?
Yes - free forever, no credit card. You only move to a paid plan when you need more agents, environments, jobs or seats, or schedules that run more often. Every plan has every feature.
Can my whole team use it?
Yes. Invite teammates and give each a role per environment (Viewer, Operator, Editor, Admin). It's one workspace and one bill, and your plan caps how many people can run or change things; Viewers are free and unlimited. Seats per plan: Free 2, Starter 3, Team 10, Business 30, Enterprise unlimited.
Which operating systems are supported?
The agent runs on Windows. It executes whatever commands you'd normally run there - cmd, PowerShell, Python, and so on.
Can I cancel anytime?
Yes, from the Billing page. Downgrading or canceling never deletes anything you've already created.
Install an agent. Create a job. Watch it run.
Free forever, no card. Every plan has every feature; you only pay for scale.