IN Brief:
- Datamyne covers 230 markets and receives more than 500 million shipment records annually.
- Users can ask natural language questions and inspect the shipment records supporting generated findings.
- Descartes claims research and analysis time can fall by up to 90%, depending on user experience and query complexity.
Descartes Systems Group has added a conversational artificial intelligence agent to its Datamyne global trade intelligence platform, giving sourcing and supply chain teams a natural language route into a dataset spanning 230 markets.
The Descartes Datamyne AI Agent is embedded within the existing platform and draws on more than 500 million shipment records added annually. Users can ask questions in ordinary language and receive contextual answers supported by dynamic visualisations and the underlying shipment records used to produce them.
The tool is aimed at sourcing, sales, supply chain, and market intelligence teams that would otherwise build filters, use Boolean logic, or work through large sets of trade records manually. Descartes says users can refine a question, investigate related companies or commodities, and move from a broad market query into the shipment activity behind a particular supplier, buyer, or trade lane.
That ability to trace an answer back to the source data is more important than the conversational interface alone. Trade intelligence is useful only when an analyst can test whether a finding is supported by real shipment activity, particularly when the result may inform supplier selection, market entry, inventory planning, or a change in sourcing strategy.
Descartes says the AI Agent can reduce global trade research and analysis time by up to 90%, depending on the user’s experience and the complexity of the query. The figure is a supplier claim rather than an independent benchmark, and the practical test will be whether teams can reproduce that saving across routine work without sacrificing verification.
The platform is designed to analyse observed trade behaviour rather than rely on company descriptions, supplier profiles, or general web content. That allows users to examine changes in product flows, shipment volumes, suppliers, buyers, and market demand using the records collected within Datamyne.
For procurement teams, the appeal is straightforward. A category manager assessing an unfamiliar market may need to identify active suppliers, compare export volumes, check whether a supplier serves particular destinations, and establish whether volumes are rising or declining. Each of those questions can require several database searches when the analyst has to construct the filters manually.
A natural language layer can shorten that mechanical work, but it does not remove the need to define the question correctly. Product categories can sit across several tariff codes, legal entities may trade under different names, and shipment records vary in detail between jurisdictions. A faster query against the wrong classification still produces a weak answer.
The same caution applies to supplier discovery. A shipment history can show that goods moved, how often they moved, and where they were sent, but it does not prove manufacturing capability, financial strength, quality performance, regulatory compliance, or available capacity. Those checks remain separate parts of supplier qualification.
Datamyne’s value therefore lies in narrowing the field and exposing patterns that deserve further investigation. If a business can establish that a supplier has recently increased exports of a relevant commodity, or that a competing buyer is sourcing more heavily from a particular country, the next stage is to test the commercial explanation rather than treat the movement itself as the conclusion.
That distinction is particularly useful when trade conditions change quickly. Tariffs, sanctions, port disruption, supplier failures, and shifts in regional demand can alter trading patterns before a conventional supplier review is completed. Access to recent shipment activity gives sourcing teams another way to test whether a change is isolated or visible across the market.
Keeping the AI function inside the trade intelligence platform also avoids moving sensitive analysis into a separate general-purpose tool. The query, visualisation, shipment records, and follow-up investigation remain within the same environment, which gives users a clearer audit trail for how an answer was reached.
There is still a risk that conversational access encourages users to accept a neatly written response too quickly. Descartes has addressed part of that problem by exposing the records behind the analysis, but responsibility for checking the scope, data quality, and commercial interpretation remains with the user.
The system is best viewed as an interface for a large trade dataset rather than a substitute for procurement judgement. Its strongest use case is reducing the time spent building and rerunning searches while preserving access to the evidence needed to challenge the result.
With more than 500 million shipment records being added each year, the volume of data is already beyond what most teams could inspect manually. The value of the AI Agent will depend on whether it helps those teams reach a smaller number of defensible decisions more quickly, rather than simply generating more analysis from the same underlying records.
