Platform capability · in mPorts OMS
Ask what happened. Know why. Take action.
Ask in ordinary business language why an order cannot ship, why a channel shows stock the warehouse does not have, or what the warehouse billed. AI in mPorts OMS reads the same orders, products, inventory and warehouse billing your team works from, names the cause, recommends the next step and shows why — and once a person confirms, runs it through the same steps a person would use, confirms the result from the connected system’s own record, and keeps the run.
One path, every question
The same steps whether the question is about an order, a channel or a bill — and the change is confirmed where it went before it is called done.
- 01
Ask
What happened? Typed in business language, against the orders, products or costs the operator is looking at.
- 02
Investigate
The relevant operational records are checked at once: order state, warehouse acknowledgment, inventory position, tracking, what the warehouse billed.
- 03
Explain
The cause is named — where the order stopped, which check blocks it, which fee and warehouse a cost sits in — with the records it looked at shown.
- 04
Recommend
The next step, and why: change the warehouse, resubmit, requote the shipment, resend to the retailer.
- 05
Act
A person confirms. The step runs through the same path a person’s change would take, into the connected systems.
- 06
Verify
The connected system’s own record is checked — does the warehouse hold the order, did the retailer receive the update. Verified, stale or unknown; unknown is never shown as success. The change-warehouse and change-objective playbooks end by re-running the diagnosis.
- 07
Record
The question, the recommendation, the confirmation, the change and its verification are kept as one run, beside the order — and the record says the AI proposed and a person executed.
What this looks like in the product
The product’s navigation and field names, as a diagram.
AI alongside an order
Asked in business language, answered from what is recorded
- Ask
- Investigate
- Explain
- Recommend
- Confirm
- Act
AI follows one path, and a person stays in it wherever a change would be made.
One question, worked end to end
An operator asks: why are my orders from this retailer stuck today?
-
Ask
The question is typed in ordinary business language, against the orders the operator is looking at.
-
Investigate
The AI reads the connected records: order state, warehouse acknowledgment, inventory position and tracking.
-
Explain
The stuck orders come back grouped by where each one stopped, with what it looked at shown.
-
Recommend
For each group, the step that fixes it, prefilled and ready to run.
-
Act
A person confirms each step that changes an order; the approved step runs through the same steps the operator would have used.
-
Record
The question, the recommendation, the confirmation and the change are kept as one run.
What the assistant reaches today
Two lists, kept apart on purpose: what the assistant reads, and what it changes after a person confirms.
Reads today
- Orders — find, list and open them, including everything needing attention
- Why an order cannot ship, with the reason and the fixes that clear it
- Products and inventory balances
- What the warehouse billed, by fee type and warehouse
- Which report runs finished
Acts today, after a person confirms
- Change the warehouse for an order
- Change the routing objective or the label owner
- Refresh the plan and submit to the warehouse
- Requote the shipment
- Resend the order to the retailer
-
Assistant
Ask in business language, from the order, the product or the cost you are looking at.
-
Agents and policies
Agents are set up per customer with the tools each may use and the rules it runs under.
-
Approvals
A change waits in an approval queue until a person with permission approves it; the requester cannot approve their own.
-
Run history
Every run kept — the question, the records read, the recommendation, the decision and the change — and a page your team can open to see what the AI proposed and did.
-
Verification
After a change reaches a warehouse or retailer, mPorts confirms the result from that system’s own record and marks it verified, stale or unknown. Unknown is never shown as success.
-
Attribution
When the AI proposed and a person executed, the audit record links the action to the proposal and keeps the person as the actor.
A chat window beside the screen, or AI inside the operation
Most “AI for commerce” is a copilot: it reads what is on the screen and drafts an answer. AI operations in mPorts OMS is built differently, because operating is different from describing.
A copilot on top
- Reads whatever the screen shows
- Writes an answer; a person goes elsewhere to act
- No fixed set of things it may change
- The change is done when the button is pressed
- What happened lives in a chat history
AI operations in mPorts OMS
- Reads the same orders, inventory, fulfillment and billing records the team works from
- Recommends, and after a person confirms, runs the step through the same workflow
- A reviewed set of capabilities; writes denied unless policy allows
- The change is done when the connected system’s own record confirms it
- What happened is one run on the order, attributed and kept
Current boundaries
The AI works from the operational information the system already holds: orders, products, inventory, fulfillment, shipments, what the warehouse billed and what was promised. It does not read your documents, policies or standard operating procedures. Product launch readiness, stock age and per-SKU cost are screens you open in mPorts OMS today; the assistant does not reach them yet. Agents are configured per customer; no specialized operator is sold as a product.
Where the operating model goes next
Because every operator would share the same records and the same controls, one AI capability can extend across the work an operations team already does. These are illustrations of that direction.
- Illustrative
Order operations
“What is stuck today, and why?”
Find the orders that stopped moving and trace each one to the stage it stopped at.
- Illustrative
Inventory
“Where are we at risk of overselling?”
Spot the channels selling against stock that is already committed somewhere else.
- Illustrative
Fulfillment
“Which orders should be rerouted?”
When a plan breaks, find the location that can still meet the promise.
- Illustrative
Product launch
“What is blocking this SKU from going live?”
Work back from the channel’s rejection to the attribute or feed that caused it.
- Illustrative
Reconciliation
“Where do two systems disagree?”
Find the records that differ between the warehouse, the channel and the OMS.
- Illustrative
Retailer compliance
“What does this retailer require that we have not done?”
Check an order against the retailer’s rules before the chargeback arrives.
One path for every operator — the way AI in mPorts OMS works today
- Ask
- Investigate
- Explain
- Recommend
- Act
- Record
The six areas illustrate the direction, not products you can buy today.
Next: Operational Memory
Everything above is written down today: each run with the question, the records read, the recommendation, the decision, the change and whether it was verified, attributed to the AI that proposed and the person who executed. mPorts is building toward Operational Memory: a record of what happened before, why a decision was made, what action was taken and what resulted, so the next decision can start from prior operational experience. It is a stated direction, not a shipped capability; today AI in mPorts does not change how it behaves based on what happened before.
Questions buyers ask
What can AI do inside mPorts OMS?
It answers questions about live orders, products, inventory and warehouse billing from the records mPorts OMS already holds, explains why an order cannot ship, recommends the step that clears it, and — after a person confirms — changes the warehouse, resubmits, requotes a shipment or resends to the retailer through the same steps your team uses. Every run is recorded.
Can we restrict what the AI is allowed to do?
Yes. Each agent has the tools it may use and the rules it runs under, and a change waits in an approval queue for a person with permission before it runs.
Which questions can the assistant answer today?
Why an order cannot ship; which orders need attention for a retailer or warehouse; what a product’s inventory balance is; what the warehouse billed by fee type and warehouse; which reports have run. Launch readiness, stock age and per-SKU cost are screens you open today.
Ask us what your team asks each other every morning.
Tell us the questions someone answers by opening four systems today. We will show you which ones the assistant answers now, and how a confirmed action runs.