The shipping season does not wait for a report. When a vessel is booked and the question is whether the stock exists on site to fill it, the answer needs to be now, not Thursday. And in most commodity export businesses, Thursday is when the report arrives.
BSM Global’s Buy Sell Move platform holds everything an exporter needs to answer that question: contracts, orders, packing, containers, vessels, markets, sellers, sites, freight rates. Getting it out in the shape you need at the moment you need it is a different matter. This post is about BSM Global reporting through an AI assistant that answers from the live system in plain English, and about one counting problem that shows why the connection is the easy part.
BSM Global makes Global Trade Management software for the commodity industry. It handles contracts, orders, shipment planning and execution, commercial documentation, compliance and tariffs, and it talks to shipping lines, customs agents, banks, terminals, government authorities and transport companies. It sits alongside ERPs like SAP, Oracle and Business Central.
Its customers are commodity traders and exporters. In WA that means hay, grain and fodder going out in containers to Japan, Korea and the Middle East, plus dairy, import and retail businesses further afield. The reader we have in mind runs export operations or finance at one of them. Their world is volume, timing and cost, and their frustration is that the system knows the answer and will not tell them quickly.
To be clear before we go further: we are not affiliated with, endorsed by, or a partner of BSM Global. We are a Perth managed IT provider, and this is an AI assistant we build and run that connects to a client’s BSM environment with their authorisation.
We connect Claude, Anthropic’s AI model, to your BSM environment through an open standard called the Model Context Protocol. You ask in plain English, in Microsoft Teams or the Claude app, and Claude answers from the live data as you. With your approval it prepares changes and drafts documents.
It reads the live system, not a nightly copy. It acts as the person asking, so permissions follow the person and every question is logged. It runs where the operations team already is. The managed AI service around it handles the connector, the access model, the logging, and the tuning against your data, which for reasons we will get to is most of the work.
“How many containers went to Japan this financial year?”
“What has shipped this month versus the same month last year?”
“Break our volume down by market and by shipping line.”
There is a detail here worth being precise about, because it is the kind of thing that separates a tool that knows the industry from one that does not. Containers are counted by movement, not by container number. The same physical box is leased and reused across unrelated orders through a season, so counting unique container numbers badly undercounts what you actually shipped. Ask the question and the assistant uses the right basis, and tells you which basis it used. A generic tool would count the numbers, give you a figure a third too low, and look perfectly reasonable doing it.
“What is packed but not yet shipped?”
“Which orders have an ETD in the next fortnight?”
“Show me everything open for this seller.”
“What is sitting on site at each location right now?”
“Which orders are short against what was contracted?”
That second one, with the season closing, is the question that decides whether a vessel sails full.
“Who are our biggest sellers by volume this season?”
“How has the Japan market moved over three years?”
“What are we paying per container on this lane?”
“How has freight cost moved against volume this year?”
“Export that to a spreadsheet.”
Results come back as a table you can sort and chart in the conversation, and download as a workbook when you want to keep it or send it on. For an operations manager who lives in Excel, that is the difference between a curiosity and a tool.
“Update the ETD on these orders to the new vessel date.”
“Flag these containers as loaded.”
“Draft the shipment summary for the customer.”
Changes are prepared for review before they are recorded. The vessel date change is shown across every affected order before it is written. The summary comes back as a draft. A person confirms. In a business where a wrong ETD propagates to the customs agent and the customer within the hour, that pause is not bureaucracy.
BSM is explicitly built to sit next to other systems. It already talks to shipping lines, customs agents, banks, terminals and ERPs. That makes the cross-system questions particularly natural here, and particularly valuable, because the gaps between those systems are where money and paperwork go missing.
“Do this season’s shipped volumes reconcile with the revenue recognised in the ERP?”
“Which shipped orders have not been invoiced yet?”
“Compare freight cost per container against what has been posted to the general ledger.”
Shipped but not invoiced is money sitting on the wharf. We covered the ERP side of this in our post on a Business Central AI assistant, and the pairing with BSM is exactly the kind of question neither system answers alone.
“What did we actually pay in freight on this lane versus what was quoted?”
“Which shipping line invoices do not match the containers we have recorded?”
“Pull together the documentation for this shipment and tell me what is missing before the vessel sails.”
“The customs agent has emailed about this consignment. Find the order, the containers and the paperwork, and draft the reply.”
Documentation completeness before a sailing is an expensive problem in this industry. A container that misses a vessel because a phytosanitary certificate was not attached costs real money in demurrage, rebooking and customer goodwill. It is exactly the check nobody has time to run manually across every consignment, and exactly the check an assistant with the order, the inbox and the document library in view can run every morning.
“Which growers have delivered against contract and which are short with the season closing?”
BSM alongside Microsoft 365 (email, Teams, SharePoint), Business Central and Xero runs today. If your ERP is SAP or Oracle, or your growers are managed in a separate system, that is a build rather than a blocker.
We have written a series of these posts, one per business system, and the same paragraph appears in every one because it is the most important thing we have learnt doing this work. BSM is the clearest example of it, so this is the long version.
The genuinely valuable part of connecting AI to a business system is not the connection. Anyone can do that now. It is knowing where the data misleads you.
Every trade data warehouse has reporting views that quietly exclude rows. Joins that silently drop records when a lookup is missing. Tables that look like reference data and have not been maintained since the system went in. Status codes that half the team uses one way. And the container-counting problem above, where the obvious method is wrong by a third and nothing about the output tells you so.
Point a generic AI at a warehouse like that and ask how many containers you shipped to Japan, and you will get a confident, specific, wrong number. It will be formatted nicely. It will look completely reasonable. Nobody will catch it, because the whole point was not having to check, and the number will end up in a board paper.
Making an integration trustworthy means measuring the data against itself: running the same question three ways and seeing where they disagree. Writing down where the data lies. Building those traps into how the assistant answers, so that when it gives you a container count it says “by movement, excluding cancelled orders” without being asked. That is what a client is paying an integration specialist for. It is precisely what a self-serve connector does not give you, because a self-serve connector does not know your business, your season or your reused container leases.
We think this is the honest sales pitch for the whole category. Not “we can connect your system to AI”. Everyone can. “We know where your data lies, and we built the assistant so it tells you.”
Commodity exporters and traders running BSM Global, in WA mostly hay, fodder and grain, with an operations team that lives in the platform and a finance team that reconciles it against the ERP. Businesses where a vessel missed or a container miscounted costs more than the software does.
We are a Perth managed IT provider, we support agribusiness and resources clients across WA, and we deliver this hands-on. BSM’s customers are spread across Australia and the connector works anywhere. If you are running BSM in South Australia or Victoria and these are your questions, talk to us and we will say plainly whether we can deliver it well.
Count your containers two ways. Pull this season’s shipments and count unique container numbers, then count movements. If the two figures differ by more than you expected, you have just found the first place your data misleads you, and every report that used the first method has been wrong. That is worth knowing before anyone connects anything.
List the checks you wish someone ran every morning. Documentation complete for every sailing in the next fortnight. Packed but not shipped. Shipped but not invoiced. Growers short against contract. That list is the brief.
Watch one of them answered live. We have a working demonstration against a realistic commodity export dataset. Bring your question, watch the answer come back with the basis it used and the records behind it. Twenty minutes, no commitment. Contact us on 1300 EPIC IT or through the contact page.
Yes. A BSM Global AI assistant connects an AI model such as Claude to your live Buy Sell Move environment so you can ask about shipments, orders, stock, sellers and freight in plain English. Ours runs in Microsoft Teams, respects each user’s permissions, logs every question, and prepares changes for a person to approve. We are not affiliated with BSM Global.
By movement, not by container number. The same physical container is leased and reused across unrelated orders through a season, so counting unique numbers undercounts what was shipped. A reporting tool for commodity export should use movements as the basis and say so. Ours does both.
Yes. Any answer that comes back as a table can be sorted and charted in the conversation and downloaded as an Excel workbook. Ask the question, then say “export that to a spreadsheet”.
With BSM, your email and your document library connected, yes. Ask it to pull together the documentation for a shipment and report what is missing, or run that check across every sailing in the next fortnight. A container that misses a vessel for a missing certificate is one of the more expensive avoidable costs in commodity export.
Business Central and Xero alongside BSM run today, which lets you ask whether shipped volumes reconcile with recognised revenue or which shipped orders have not been invoiced. SAP and Oracle are a build rather than a blocker. Once both systems are connected, questions that span them work without a project.
No. We are a Perth managed IT provider with no affiliation to BSM Global. This is an AI assistant we build and operate that connects to a client’s BSM environment using their own authorisation, alongside their other business systems.
