Descartes ships Datamyne AI Agent for conversational trade intelligence

Descartes Systems Group issued a GlobeNewswire announcing the Descartes Datamyne AI Agent, a conversational capability inside Datamyne that answers a trade question in plain language and shows the shipment records beside the answer, with a company claim of up to 90% less research and analysis time.

TECH desk — someone sourcing a part used to build filters and export tables before they could see who started shipping. A chat that answers in a sentence and leaves the shipment rows on the screen is how that person checks the answer instead of trusting the paragraph.

What the lede says the announcement is. Descartes Systems Group (Nasdaq:DSGX) (TSX:DSG), called on the page the global leader in uniting logistics-intensive businesses in commerce, announced the Descartes Datamyne AI Agent. The page prints a trademark mark on Datamyne. Global leader is the company’s adjective. Nasdaq:DSGX and TSX:DSG are the tickers the release prints. They are not a price, and they are not advice to buy or sell. The agent is a new conversational AI capability that helps sourcing, sales, supply chain, and market intelligence teams reduce global trade data research and analysis time by up to 90%, depending on the user’s experience and query complexity. Up to sits in front of 90%. Ninety percent is about nine tenths, if the claim holds. That translation is this desk’s. The percentage is the company’s, and the release ties it to who is asking and how hard the question is. This desk did not time a search.

Where the agent sits, and what comes back. Embedded in the Descartes Datamyne global trade intelligence solution, the AI Agent lets users ask questions in natural language and returns contextual answers in seconds, supported by dynamic visualizations and the underlying shipment records. Natural language, here, means a sentence a person would say, not a form full of filters. A visualization is a chart or a table the screen draws from the answer. Those glosses are this desk’s. In seconds is the release’s clock. The subhead says the agent draws on data spanning 230 markets with more than 500 million shipment records annually to deliver traceable insights in seconds. The body says Descartes Datamyne covers 230 markets with more than 500 million shipment records added annually. Spanning, and added, are two verbs. The subhead says the data the agent draws on. The body says records added each year. Do not collapse them into one total of 500 million rows sitting in a warehouse. More than 500 million is the company’s floor. 230 is the company’s market count. Traceable, here, means the answer points at the records behind it. That gloss is this desk’s. The release prints the word.

Brian Hodgson, as a quote, not as a study. Brian Hodgson, General Manager, Trade Intelligence at Descartes, said geopolitical disruption, tariff volatility, and shifting sourcing patterns are making trade decisions more complex and time-sensitive. He said that by making trusted trade intelligence more accessible, the agent helps organizations understand where suppliers are gaining or losing activity, how product flows are changing, where market demand is emerging, and how competitors are adapting. He said users move quickly from a business question to validated insights on trends, risks, and opportunities while retaining visibility into the records behind the analysis. Validated, in that sentence, is his word for an insight a person can check against the records. That gloss is this desk’s. Those are his sentences on the release. This desk did not interview him.

What the release says a person can do with the records, and only those jobs. The agent makes the import and export data easier to access, helping users identify suppliers, monitor competitors, evaluate trade risks, develop sales opportunities, and understand changing markets. In seconds, it responds to natural-language questions with relevant data, visualizations, and contextual insights. The release then lists five helps. A person can ask complex trade questions in natural language without building filters, using Boolean logic, or mastering database query structures. Boolean logic, here, is the AND and OR a search box uses to combine rules. That gloss is this desk’s. The release prints the phrase. A person can validate AI-generated findings through dynamically updated data views, visualizations, and drill-down access to supporting shipment records. Drill-down means opening the rows under the answer. That gloss is this desk’s. A person can explore inside the same environment by refining questions and investigating related companies, commodities, shipments, and trade relationships. A person can analyze actual trade behavior using observed shipment activity to understand changes in suppliers, buyers, product flows, and market demand, rather than company descriptions, self-reported supplier profiles, or general web content. Observed shipment activity is the release’s contrast with a company’s own write-up of itself. A person can improve supply chain planning with decision-ready summaries that highlight notable trends, anomalies, sourcing opportunities, and other patterns across large volumes of trade records. An anomaly is a number that does not fit the usual pattern. That gloss is this desk’s. These five helps are the release’s. This desk did not ask the agent a question.

Ken Wood, as a quote. Ken Wood, EVP Product Management at Descartes, said AI is most useful when it combines trusted, domain-specific data with a clear connection to the decisions customers need to make. Domain-specific, here, means the data are about this job, trade shipments, not a general chat model’s web reading. That gloss is this desk’s. He said the agent analyzes shipment-level data and presents findings alongside the records and visual evidence supporting them, so users can verify conclusions, refine their questions, and continue exploring without leaving the platform. He said embedding the agent directly into Descartes Datamyne gives customers a more transparent and intuitive way to accelerate decision-making and respond more effectively to changing global trade conditions. Transparent, intuitive, and more effectively are his words. They are not the 90% line. Those are his sentences on the release. This desk did not interview him.

What the about box says the company is, and the caution that sits under it. Descartes says it powers more responsive, efficient, secure, and sustainable international and domestic supply chains by uniting logistics-intensive businesses on its Global Logistics Network, shortened to GLN. Shippers, carriers, and logistics service providers connect and collaborate on the GLN, using technology, data, and AI to manage last-mile deliveries, domestic and international shipments, transportation rating and payment, global trade research, customs compliance, and a variety of regulatory processes. A last-mile delivery is the final leg to the door. That gloss is this desk’s. The site it names is www.descartes.com. The release also says to connect on LinkedIn and X. The global media contact on the StockTitan reprint is Cara Strohack, telephone 226-750-8050, cstrohack@descartes.com. Quiver redacts that address as an email placeholder. This desk did not email it. The release then prints a cautionary statement on forward-looking statements, including that results may differ and that the trading price of the common shares could decline. That block is a securities warning. It is not a product measurement. This filing is not investment advice.

What the July product page adds, and what it does not let you file as the wire. The Datamyne knowledge-center page, “The Future of Global Trade Intelligence: Introducing the Descartes Datamyne AI Agent,” is by Descartes Datamyne and dated Jul 24, 2026. That is two months before this wire. The page this desk read does not print 24 Sep 2026, does not print 06:45 ET, does not print up to 90%, does not print 230 markets, and does not print 500 million records. It says a person can ask the agent to run a complex query and get a nearly instantaneous, data-rich response. Nearly instantaneous is that page’s clock. In seconds is the wire’s. Do not collapse them into a stopwatch this desk held. The page’s example questions are which companies have recently started exporting automotive parts from Mexico, who the fastest-growing suppliers of electronics into the United States are, which importers have diversified sourcing away from China over the past year, and what emerging trade trends to monitor. Those examples are the July page’s. They are not the aluminum question on the still. The page says the agent is backed by shipment-level import and export data, that it updates visualizations inside the platform, and that it can generate summaries, reports, and alerts. It calls Datamyne the world’s most comprehensive global trade intelligence platform. World’s most comprehensive is that page’s claim. The wire’s adjective is global leader, in a different sentence. The page prints servicedesk@descartes.com and the telephone 833.262.2315. Those are the July page’s. They are not the media contact on the wire. This desk did not call either number.

Accuracy on the product date. The 24 Sep wire says Descartes announced the agent. It does not say this is the first day the product existed, and it does not print a first-ship date. The July 24 page already introduces the same agent and shows the assistant chat. This filing treats the 24 Sep GlobeNewswire as the same-day company primary for the up-to-90% announcement and the 230-market, 500-million-record sentences. It does not treat 24 Sep 2026 as a brand-new first-ever product date beyond the wire’s own wording. This desk did not treat the July page’s claims as today’s measured result.

What the card shows, and what the words do not. The card is the Datamyne assistant chat from the company’s Fig-1 on that July page. The assistant says “Hi I am your Datamyne assistant. How can I help you today?” The question on the still is “Who are the fastest-growing suppliers of aluminum products?” The assistant asks which period to use, and offers this year versus last year, the last 12 months versus the previous 12 months, or a custom date range. The reply on the still is the last 12 months versus the previous 12 months. The answer lists five names: AUGUST AMBER HK LIMITED (CN), up from 0 to 3,807 records; AUGUST AMBER HK LIMITED (HK), up from 0 to 1,094.58 records; ZHEJIANG KANG SHENG COMPANY LIMITED (CN), up from 0 to 739.15 records; NOVELIS KOREA LIMITED (KR), up 546.0 records, about +49.6%; AMAG ROLLING GMBH (AT), up 161.47 records, about +12.9%. A closing line on the still says most of the growth is coming from new or newly active shippers, while several established suppliers are flat or slightly down. The box at the bottom says “Ask about trade data...” Those names, the decimals, the percents, and the period choice are the graphic’s. The 24 Sep release does not print them. The July page’s example question is electronics into the United States, not aluminum. Do not treat the still as a supplier ranking this desk calculated, and do not treat 3,807 records as the wire’s 90% claim. The card has no date text.

Plain English for the rest of the card. Datamyne is Descartes’ trade-intelligence product. The agent is a chat inside that product. A person asks in a sentence. The answer is supposed to come back with a chart and the shipment rows, so the person can check it. Up to 90% is the company’s claim for less research and analysis time, and it depends on experience and how hard the question is. 230 markets, and more than 500 million shipment records added annually, are the company’s coverage lines. The subhead’s “spanning” and the body’s “added” are two sentences. Boolean logic is AND and OR in a search. A drill-down is opening the rows under the answer. GLN is the Global Logistics Network, the about box’s name for the network shippers and carriers share. 06:45 ET is 6:45 a.m. Eastern and 10:45 a.m. UTC. 10:50 a.m. UTC is Quiver’s clock, five minutes later. Jul 24, 2026 is the product page’s date. Nasdaq:DSGX and TSX:DSG are the tickers on the wire. Brian Hodgson is general manager of trade intelligence. Ken Wood is executive vice president of product management. Cara Strohack is the media contact on the StockTitan reprint. This filing is the 24 Sep announcement.

PRIMARY here: Descartes’ 24 Sep 2026 GlobeNewswire, release 3368165, header September 24, 2026 06:45 ET on the snippet this desk could read, datelined London and Atlanta on the StockTitan and Quiver reprints — Tier A PRIMARY, the company’s own announcement. globenewswire.com did not return HTML on the fetches this desk made. The lede, the up-to-90% claim, the embedded-in-Datamyne sentence, the in-seconds sentence, the subhead’s 230 markets and more-than-500-million annual records, the body’s added-annually sentence, the Hodgson quote, the five helps, the Wood quote, the about box, the GLN, the Cara Strohack contact as StockTitan prints it, and the forward-looking caution are that release’s, as those reprints carry them. The July 24 product page is prior company context, including the example questions and the world’s-most-comprehensive line. The aluminum names, the record counts, and the period choice are the Fig-1 still’s. NOT claimed: that this desk loaded the GlobeNewswire HTML, that 24 Sep 2026 is the first day the agent existed, that up to 90% was timed here, that 230 or 500 million was audited, that spanning and added are the same total, that the aluminum list is a ranking this desk ran, that 1,094.58 or 739.15 was rounded, a named customer, a dollar price, a stock tip, or investment advice. Distinct from the already-filed autonomyai-discover-mode, order-purchasing-copilot, and dun-bradstreet-db-ai.