Four things to decide, not four hundred to check.
Every night Planster re-checks every SKU, order, and channel against your plan. Every morning you get a ranked email of what needs a decision, with the purchase orders already written. You approve them in Planster.
What it is reading: one demand forecast built across DTC, Amazon, retail, and wholesale from 150+ connected systems, netted against what is on hand, on order, and already committed. The agent does not have its own numbers. It has yours.
Ask anything — every answer carries receipts
The chat answers inventory, purchasing, and demand questions with provenance or not at all: "4,212 units on hand · as of last night’s sync," linked to the screen showing the same number. A figure that can’t carry a link doesn’t get stated.
A morning digest, not a dashboard to check
The subject line carries the headline — "2 SKUs need POs this week, 1 stockout risk" — so triage happens in your inbox. Items are ranked and capped. Quiet days say all clear. You only log in when something needs you.
The orders are already written
Every item that needs a purchase order has one written against it, waiting in the app. You change the quantities you want changed and approve. Nothing reaches a supplier before that.
It sees what you see
The agent reads exactly what the user it works for can see — no more. Same numbers, same permissions, same moment. It works from your live data, not a copy of it.
When it’s wrong, it says so
Every message carries a flag. Flag one and it goes to a review log. Silence any check you don’t want to hear about again. When a flagged issue gets fixed, the next morning’s digest tells you what caused it. Errors get a paper trail and a visible ending.
Hard limits, published up front
It reads everything you can see. It writes draft purchase orders and report definitions. That’s the list. It cannot promote a plan, approve an order, send anything to a supplier, or delete. There is no setting that changes this, which means the worst case is a draft you delete.
The loop, end to end
At 7:00 AM the digest flags PASTE-MINT: 18 days of cover left, stockout projected for late August. She asks the agent what’s driving it and gets the answer with a link to the screen that proves it — Amazon velocity up 40% for three straight weeks. At her desk she opens Planster and the order is already written. She changes it to 10,000 units and approves. Underneath it, a note that the Albertsons deal was marked won but never promoted, with a link to the promote screen. Two decisions, ten minutes.
What it works on
Every capability below carries its own section on what the agent does with it — and what it cannot do.
Know what’s coming
One forecast across every channel, and every number on it accountable.
Buy the right thing
What to order, when, how much, and from whom.
See what it costs first
Scenarios, retail deals, and the one plan sales and ops both read.
Common questions
Can the agent order inventory on its own?
No. The agent writes purchase orders as drafts. Every one requires your approval inside Planster before anything reaches a supplier, and there is no setting that changes this.
Is this replacing my planner?
No. It replaces the part of their morning spent hunting through screens to find out what changed overnight. The agent watches, ranks, and writes; your team judges, decides, and approves. Every gate a human passes through today still applies.
What data can the agent see?
Exactly what the user it works for can see — no more. Its permissions are read-everything, write-drafts-only, and every number it reports links back to the in-app screen showing the same figure.
What happens when it gets something wrong?
Every message carries a flag. Flags go to a review log, you can silence any check yourself, and when a flagged issue is fixed, your next digest tells you what the cause was. Errors get a paper trail and a visible ending.
See what last night would have caught
Connect your data and get your first ranked list — with the math already done and nothing committed without you.