Automate the admin, not the judgement
TaskIt automates the parts of task management that are pure clerical overhead — creating the repeating task, telling the right person, sending the summary, retrying the message that failed — and leaves the decisions to people.
No separate automation tier. It is included on every plan term.
The problem
Most of the effort in "managing tasks" is not management
Count what a supervisor actually does in a day and a large share of it is transcription: retyping the same weekly task, telling someone it exists, writing the evening summary, chasing whether a message was seen.
- The same fifteen recurring tasks are recreated by hand every week, slightly differently each time.
- Each assignment needs a second message to tell the person about it, because the system they were assigned in is not the app they read.
- The daily summary is typed out by one person, and it stops when they are away.
- A failed notification is discovered when the work is not done, not when the message failed.
- Everything above scales linearly with headcount, which is why it is usually the first thing to be dropped.
How TaskIt handles it
Four kinds of automation, all of them boring on purpose
Automation here means the system doing what it can prove it should do. There is no rule engine guessing at intent.
- Scheduled task generation: recurring schedules create real tasks on a daily, weekly, monthly, yearly or custom cadence, in a chosen time zone.
- Event-driven notifications: task creation, handover and completion each trigger in-app and optional WhatsApp messages, switchable per event.
- Scheduled reporting: daily, weekly and monthly WhatsApp reports at a set time, with the statistics window following the cadence.
- Reliable delivery: every outbound message goes through a persistent, priority-ordered queue with retries, exponential backoff, de-duplication and crash recovery.
- Automatic attribution: a task created without a department inherits its assignee’s department, so reporting stays accurate without anyone remembering a field.
What you get
The routine runs without a person behind it
A schedule holds the cadence, the template and the assignment. Leave, turnover and a busy week stop being reasons the recurring work did not get raised.
Assignment and notification are one action
Creating a task is what tells the assignee about it. There is no second message to send and no gap between the two where work goes missing.
Reporting arrives instead of being requested
The daily, weekly or monthly summary is generated from records and sent on schedule, so the management rhythm does not depend on one person’s diary.
Failures are visible, not silent
Retries are automatic and bounded, execution history records every attempted recurring run, and delivery history records every message — including the ones that failed and why.
How it works
In the order a person using it meets them.
- 1
Automate what repeats
Turn each routine — opening checks, weekly reconciliation, monthly filing, annual renewal — into a schedule. Preview the occurrences before committing.
- 2
Decide what people should be told
Switch creation, handover and completion notifications on or off, and choose whether each goes to the person, a group, or both.
- 3
Set the reporting rhythm
Pick a cadence and a time. Preview it, test-send it, then leave it — the scheduler evaluates each report in its own time zone and claims each run atomically so it cannot double-send.
- 4
Check the evidence occasionally
Recurring execution history and WhatsApp delivery history are where you confirm that the automation is doing what you think it is.
What a week costs before and after
A 20-person operations team, counting only the clerical part.
- 1
Recreating recurring tasks
Fifteen routines re-raised by hand each week becomes five schedules that raise them, in their own time zone, whether or not anyone is in.
- 2
Telling people
A second message per assignment becomes nothing at all — creating the task is what notifies the assignee, the group, or both.
- 3
The evening summary
One person typing a recap becomes a scheduled report with today’s assignments, progress, completions and anything blocked.
- 4
Chasing delivery
Asking "did you see it?" becomes glancing at the delivery history, which already shows sent, retried or failed.
The result
The judgement calls — who should do this, is this actually finished, why is this blocked — are exactly where they were. Only the transcription is gone.
Generation that cannot duplicate
An occurrence is claimed in a run ledger before a task is created, under a unique schedule-and-occurrence constraint. Two overlapping scheduler ticks therefore cannot create the same task twice — this is guaranteed by the database rather than by hoping the ticks do not overlap.
A queue, not a direct call
Nothing in the product calls the WhatsApp gateway directly. Every message — task events, scheduled reports, one-time codes, billing notices, test sends — is enqueued and delivered by a single background worker.
- Up to five delivery attempts with exponential backoff
- De-duplication, so a retried request does not produce two messages
- Crash recovery, so an API restart mid-broadcast resumes rather than loses
- Ordered multi-part reports: part two cannot be claimed until part one has sent
Automatic housekeeping
Storage cleanup can delete images attached to tasks completed longer ago than a retention period you choose — 30, 60, 90 days or a custom figure, disabled by default, swept nightly. Task history, comments and audit logs are never touched by it.
What TaskIt does not automate
There is no visual rule builder, no if-this-then-that engine, no conditional routing between arbitrary stages and no auto-assignment by workload or skill. Work is routed by the person raising it or by a department queue that people claim from. That is a deliberate boundary, and it is better known before a trial than discovered during one.
Frequently asked questions
No. The automation is fixed and specific: recurring task generation, event-driven notifications for creation, handover and completion, scheduled reports, a retrying delivery queue, and department inheritance on tasks. There is no if-this-then-that engine and no visual workflow designer.
No. Work is routed either to a named person or to a department queue that eligible members claim from. Auto-assignment by workload or skill is not something the product does, and a department queue is the intended answer to "I do not know who should do this".
Nothing is lost. Messages live in a database-backed queue and are delivered by a background worker with crash recovery, so a restart resumes delivery from where it stopped rather than dropping the remainder of a broadcast.
No. The scheduler claims each day’s run atomically before sending, which is what prevents a duplicate send when two ticks overlap.
Where to go next
The capabilities either side of this one, and the problems it is bought to solve.
Automate one week of clerical work
Turn your five most repeated tasks into schedules and switch on the daily report. That is usually the whole first afternoon.