An AI that knows where every pallet is — and asks before it acts.
NG Copilot answers questions about stock, orders, waves and lots and says which document each answer comes from. AI agents keep watch over the shift, dispatch, replenishment, counts and expiry dates on their own and prepare proposals, while the MCP server connects your team's AI tool to the warehouse. Every action waits for a person's confirmation.
- NG Copilot
- AI agents
- MCP server
- Answers with a document
- Actions a person confirms
Your team gets an AI that knows every order, location and lot
NG Copilot answers from your data, AI agents keep watch over the shift, dispatch, replenishment and expiry dates, and MCP connects your team's AI tool to the warehouse. Always with a document, under the user's rights and with a person's confirmation.
NG Copilot
NG Copilot knows where the goods are
Ask in your own words, in Serbian or English. Copilot goes through orders, waves, locations and goods history and answers with the document the answer comes from — and a proposed action waits for your confirmation.
- Answers with a document: wave, order, location, lot
- Shift overview and blockers before the truck
- Proposals for replenishment, a priority or a check
AI agent
AI agents for the warehouse routine
Agents keep an eye on what a shift lead would otherwise check by hand — the shift, dispatch, replenishment, counts, expiry dates and the ERP exchange — and prepare proposals that wait for a person.
- Run on a schedule or on an event
- They prepare; people decide
- The same data and rights as the user
MCP
MCP server
The open Model Context Protocol standard connects the AI tool your team already uses to NG WMS — with no separate integration per tool.
- Claude, ChatGPT and other MCP clients
- Acts on behalf of the signed-in user
- A key the administrator can revoke
See the AI at work
Moments from a warehouse day: what you ask or what an agent triggers, what the AI checks and what arrives for confirmation. The data is an example.
AI agent
The shift starts with a plan, not a walk-around
At the start of the shift, an AI agent goes through open receipts, waves in preparation, orders with blockers and pick locations below minimum, and builds a short overview for the shift lead: what is late, what is waiting for stock and where to act first.
- A shift overview with a document behind every item
- Blockers and empty pick locations before the first wave
- A priority proposal — changed only after confirmation
NG Copilot
What is late for dispatch — and what to do right now
You ask in your own words, and Copilot checks today's dispatch orders, waves, reservations and ERP stop flags — then says for each order what is holding it back and which document it knows that from.
- Orders with blockers, with the reason and the document
- Blockers kept apart from warnings
- A replenishment proposal — the task is created only after confirmation
NG Copilot
Where a lot is and who received it — in one answer
Copilot follows a lot from receipt, through delivery notes, to stock at a location, and names the document behind every statement. When a customer or an inspector asks, the answer does not wait for a custom report.
- Receipt, delivery notes and lot stock in one place
- A document behind every item, one click away
- Only the warehouses and goods owners your account may see
NG Copilot
A count variance, explained before the decision
For a variance on the worksheet, Copilot gathers the location's history — receipts, issues and transfers — and pairs a shortage with a surplus of the same item. It becomes clear faster whether it is a counting error, an unrecorded movement or a real shortage.
- Location history behind every variance
- Shortage and surplus of the same item paired
- Recount or accept — the decision stays yours
AI agent
Goods nearing expiry, in time and by location
On a schedule, an AI agent goes through lots and expiry dates in the warehouses the user may see and picks out goods that expire in the next 30 days — with the location, the goods owner and the minimum remaining shelf life for issue.
- A list by lot, location and goods owner
- A warning before goods fall below the minimum shelf life for issue
- A check proposal — the task is created only on confirmation
MCP
The AI tool your team already uses, connected to the warehouse
Through the NG WMS MCP server, Claude, ChatGPT or any other MCP-capable AI tool reads stock, orders, waves and goods history — on behalf of the signed-in user and only for the warehouses and goods owners that user may see.
- No separate integration per AI tool
- The same rights and the same trail as in NG WMS
- Proposals come back to NG WMS for confirmation
Agents keep up with the routine. Decisions stay with people.
An agent runs on a schedule or on an event, on the same orders, locations and goods history and with the same rights as a user. Anything that changes stock or tasks arrives as a proposal for confirmation.
Shift overview agent
Trigger
At the start of every shift
- Goes through open receipts, waves in preparation and orders with blockers
- Picks out pick locations below minimum and goods waiting for put-away
- Ranks the shift's priorities, with a document behind every item
Result
A shift overview for the shift lead, before the first wave
Boundary
Priorities change only on confirmation
Dispatch blocker agent
Trigger
When a wave nears its dispatch deadline
- Checks orders waiting for stock, shortages and ERP stop flags
- Keeps blockers apart from warnings in each wave
- Proposes replenishment, a priority or a check for every order
Result
A list of blockers with the reason and the document
Boundary
A person releases the wave, not the agent
Pick-face replenishment agent
Trigger
When a pick location falls below minimum
- Compares stock and reservations with the item's min/max rule
- Finds reserve source locations with available stock
- Proposes the quantity and the order of replenishment
Result
A replenishment task proposal, ready for confirmation
Boundary
The task is created only after confirmation
Count variance agent
Trigger
When a variance appears on the worksheet
- Gathers receipts, issues and transfers for the location
- Pairs a shortage with a surplus of the same item in other zones
- Proposes a recount or acceptance, with a reason
Result
An explanation next to every variance on the worksheet
Boundary
A user with the right decides and approves
Expiry and lot agent
Trigger
On a schedule, for example every Monday
- Looks for goods expiring in the next 30 days
- Checks the minimum remaining shelf life for issue, per item
- Groups the findings by location, lot and goods owner
Result
A list of goods nearing expiry and a check proposal
Boundary
A block is created only on confirmation by a user with the right
ERP exchange agent
Trigger
When a message becomes undeliverable or goes to review
- Watches the outbox and the message log in both directions
- Links the message to its document and the latest error
- Proposes a resend under the same message ID
Result
A list of hold-ups before a stock discrepancy appears
Boundary
A user with the right confirms the resend
Your team's AI tool, connected to NG WMS
Model Context Protocol is an open standard for connecting AI assistants and agents to business systems. NG WMS supports it — so the AI tool your team already uses works with your stock, orders and goods history.
AI assistants and agents
- Claude
- ChatGPT
- Your AI agents
- Other MCP clients
Model Context Protocol
MCP · NG WMS
- 01Stock by location, lot and expiry
- 02Orders and waves
- 03Receipts and announcements
- 04Goods movement history
- 05Proposed task — confirmed in NG WMS
Your rules apply
- On behalf of the signed-in user, no “super account”
- Only the warehouses and goods owners the user may see
- A key the administrator can revoke at any time
- Every call stays in the trail
One connection for every AI tool that speaks MCP — when your team changes assistant, the data, rights and trail stay the same.
Powerful AI, under your warehouse's rules
The AI can never do more than the user it acts for. Administrators decide for which warehouses and users it is enabled.
- Answers with a documentEvery answer comes with the document behind it — wave, order, location or lot, one click away. The answer is checked, not taken on trust.
- Actions only on confirmationTasks, corrections, blocks and approvals are created when a user with the right confirms them. The agent proposes; a person decides.
- Only what the user may seeCopilot, agents and MCP clients see only the warehouses, goods owners and documents the signed-in user may see in NG WMS.
- A trail of every question and confirmationQuestions, proposals and confirmed actions stay in the trail, and the document history shows who confirmed each action.
- Your warehouse rules stay yoursMinimum and maximum, the issue rule and blocks are set by the warehouse. The AI works by those rules — it neither invents nor bypasses them.
- Access and keys under controlAdministrators decide for which warehouses and users the AI is enabled and who uses MCP access, with a key they can revoke at any time.
Every AI capability, one set of rules
Copilot, AI agents and MCP clients work on the same orders, locations and goods history as NG WMS — and can never do more than the user they act for.
AI agents for the warehouse routine
Agents run on a schedule or on an event: shift overview, dispatch blockers, pick-face replenishment, count variances, expiry dates and hold-ups in the ERP exchange. The result is a proposal that waits for a person.
MCP server
The Model Context Protocol is an open standard for connecting AI assistants and agents to business systems. Through it, Claude, ChatGPT and other MCP clients read stock, orders and goods history, with no separate integration for each tool.
Questions in your own words
“Where is lot L-0915?”, “What is late for dispatch today?”, “How much of this item is in reserve?” — Copilot answers briefly and names the document, location or wave the answer comes from.
A shift overview in a minute
Open receipts, waves being picked, orders with blockers and pick locations below minimum — in one summary, at the start of a shift or before a meeting.
Blockers before the truck
Copilot points out orders waiting for stock, waves with warnings and goods that have waited too long for put-away, with the reason and the document.
Replenishment and priority proposals
When a pick location runs low, Copilot proposes replenishment from reserve and an order of work. The task is created once a user confirms it.
Count variances, explained
For every variance Copilot gathers the location’s history — receipts, issues and transfers — so the decision to recount or approve comes faster.
Lot and expiry
Which goods expire in the next 30 days, where a given lot is and who it was shipped to — the answer comes with the receipt and dispatch documents.
Same rights, same trail
Copilot, agents and every MCP client act on behalf of the signed-in user and see only the warehouses and goods owners that user may see. Questions and confirmed actions stay in the trail.
Control over AI
Administrators decide for which warehouses and users the AI is enabled and who uses MCP access, with a key that can be revoked.
From a question to a confirmed action
AI shortens the path to information; a person makes the decision.
- 01
You ask — or an agent starts on its own
In NG WMS or from an AI assistant connected through MCP, in your own words, in Serbian or English. An agent starts the same flow on a schedule or on an event.
- 02
The AI checks
It goes through orders, waves, locations and the movement history of goods, only in the warehouses your account may see.
- 03
It answers with a source
A short answer and the documents it comes from — wave, order, location or lot, one click away.
- 04
You confirm
A proposed replenishment, priority or check becomes a task only after you confirm it, with a record of who confirmed.
What it means for the business
Less searching, more work done
Shift leads, sales and management get answers from the WMS without custom reports and without calling the warehouse.
Problems surface in time
AI agents bring orders with blockers, empty pick locations and goods nearing expiry to light while there is still time to act — not when the truck is already waiting.
One standard for every AI tool
MCP is an open standard: when your team changes AI assistant, the data, permissions and trail in NG WMS stay the same.
Frequently asked questions
What is NG Copilot in NG WMS?
NG Copilot is the AI assistant built into NG WMS. It answers questions about stock, orders, waves, lots and locations, summarises the state of a shift, explains count variances and proposes replenishment and priorities — always with the document the answer comes from.
What do AI agents do in NG WMS?
AI agents run on a schedule or on an event: they prepare the shift overview, watch for dispatch blockers, propose pick-face replenishment, explain count variances and keep track of expiry dates and hold-ups in the ERP exchange. They work with the user's permissions, and anything that changes data arrives as a proposal for confirmation.
Can Copilot or the agents change stock on their own?
No. Copilot and the agents propose; corrections, replenishment tasks and approvals are created when a user with the right confirms them. Every confirmation stays in the document history.
What is MCP and what is it for?
MCP (Model Context Protocol) is an open standard for connecting AI assistants and agents to business systems. NG WMS has an MCP server, so the AI tool your team uses reads stock, orders and goods history — only with the permissions of the user working with it.
Does Copilot see every warehouse?
Copilot sees only the warehouses, goods owners and documents the signed-in user may see in NG WMS. The same applies to AI agents and to every MCP client.
How does Copilot help with stock counts?
For a variance at a location it gathers receipts, issues and transfers from the history, so it is quicker to see whether it is a counting error, an unrecorded movement or a real shortage.
Related
- Stock and replenishmentStock by location and LPN, quality blocks and replenishment that starts from a proposal.
- Cycle countA scope that fits, blind counting and a decision for every variance before it reaches the ERP.
- ERP integrationAPI, webhooks and an outbox with a log — an ERP link you can monitor and replay.
- AI in the warehouse5 min read
AI in the warehouse: what NG Copilot and MCP change in daily work
Most time in a warehouse is not spent carrying goods but looking for information. How an AI assistant that knows your orders, locations and stock history shortens that path — while people keep the decisions.
- WMS basics6 min read
What is a WMS, and when does a warehouse need more than an ERP?
A WMS runs stock by location, task and scan; the ERP keeps the books. What a warehouse management system actually does, how it differs from an ERP inventory module, and the signs that you have outgrown spreadsheets.
Want to see how this would look in your warehouse?
In a demo we walk through your receiving, picking and dispatch in NG WMS and show how NG Copilot supports your warehouse team.
- office@konissoftware.com
- Phone
- +381 66 000 977