Trimble is developing a single artificial intelligence agent designed to manage the full operational management of the logistics chain, overcoming the fragmentation of software tools that currently characterises the sector. The proposal comes from Jonah McIntire, Chief Product & Technology Officer at Trimble Transportation, who says the solution is already being tested internally within the company’s corporate networks ahead of its release to customers. The starting point is a now widespread problem: logistics coordinators still have to move between dozens of disconnected databases, management systems and tracking platforms, an activity that takes time away from higher-value work. In recent years, the sector has responded to this need with a set of specialised artificial intelligence micro-agents, each designed to automate a single repetitive task. McIntire notes, however, that this approach does not solve the problem at its root: operators continue to move from one tool to another, only now they also do so between different artificial intelligence applications.
The proposed solution consists of an advanced, single and centralised agent able to autonomously identify the specialist agents best suited to a task and retrieve the necessary data across the company’s entire information ecosystem, coordinating execution through to completion via natural-language instructions. McIntire describes it as an “intelligent and enhanced librarian”, capable of relieving users of the need to choose the right tool and manually coordinate the subsequent steps. One use case cited by the executive concerns the management of a problematic shipment: today, the process requires retrieving order history from a TMS, manually comparing spot rates stored in a spreadsheet, consulting updated market data online and finally recording a new strategy in fleet management software. A single agent, McIntire argues, could carry out the entire sequence by simultaneously orchestrating the different platforms involved.
Considerable attention is being paid to security. According to McIntire, an agent with access to company data and the ability to execute critical processes requires features that many generalist artificial intelligence tools do not provide: full audit trails, detailed records of user activity, granular permission controls and rigorous management of processing costs. The manager also notes that some systems that are still immature and lack adequate controls have displayed unexpected autonomous behaviour, such as attempts to install unauthorised software or create new API connections to bypass imposed limits in an effort to complete an assigned task. On this basis, Trimble has chosen to develop, test and validate the unified agent by using its own organisation as the first proving ground, assessing its operational effectiveness and the robustness of its security mechanisms before making it available to customers.
Antonio Illariuzzi
