How is UPM using AI?

Through a company-wide AI programme, and in the field and the mills. UPM's 2025 annual report says the programme "drives twelve high-impact transformations". It set up an AI centre of excellence and guidelines in 2025. It also ran "pilots in areas such as automated order intake, invoice handling, supplier query processes, and maintenance optimization". Outside the annual report, UPM's own stories describe more in use. A generative AI assistant is "used by hundreds of foresters" in one country, with similar tools planned elsewhere. Machine vision systems check chip flows, bale quality and wrapping in pulp operations. None of these sources puts a number on a benefit.

What the usual answer says

The top results for this question are IT partners' press releases and a page about an unrelated acronym. UPM's own story on its forestry assistant comes fourth. General descriptions list production optimisation, maintenance, forestry and supply chain.

The programme in the annual reports

The 2024 annual report describes the start. UPM had "launched an AI programme to develop a comprehensive AI strategy", developed "with contributions from more than 150 employees across multiple businesses and functions". The strategy "identifies high-impact AI transformations such as increased sales, optimised supply chains and improved processes".

The 2025 annual report reports delivery. The programme covers 12 transformations. In the report's words, it "drives twelve high-impact transformations aligned with strategic priorities", which "enable improved customer support, optimized operations and enhanced decision-making". Achievements include "establishing the UPM AI Center of Excellence to scale and sustain AI capabilities" and new AI guidelines in June. It also lists an internal chatbot and the pilots in order intake, invoices, supplier queries and maintenance. In one business, "we continued to develop our mill operations by using data and AI to optimize and improve production". The people section plans to "Pilot AI-assisted tools and chatbots to create and deliver learning content", and reports an AI learning hub.

What UPM's own stories add

A December 2025 story introduces "a generative AI companion developed by UPM that gives field forest experts instant, AI-driven answers". It helps with questions such as how many retention trees certification requires per hectare in a clearcut. The story says it "is used by hundreds of foresters", and that "UPM plans to roll out similar AI work coaches across its operations in other countries".

A June 2026 story from UPM Pulp describes "AI-driven machine vision systems" in pulp operations. They work "by evaluating pulp chip flows and bale quality, overseeing batch printing and wrapping, and monitoring unit dimensions". It adds that "AI-powered video monitoring is being piloted for wood truck drivers" to prevent fatigue-related incidents. On customers, it says "we are building customer-specific AI models that consolidate all transactional data into a single, governed interface", in tests.

These are the company's own accounts, so they show scope rather than results.

Why the annual report's pilots still matter

Three of the four named pilots are document flows: orders, invoices and supplier queries. Each has a checkable outcome, since an order line is entered correctly or it isn't. That makes them the easiest place to measure AI's accuracy. From Quarri's own work with a lumber and millwork manufacturer: a units labelling error had inflated recorded purchase spend by about a fifth. Errors like that sit in exactly these document flows, and an AI pilot that reads invoices can either find them or repeat them.

What it means for other operators

UPM's sequence has been strategy, governance and training, then pilots, then tools in the hands of field and mill staff. The field assistant is scaling to other countries without a published benefit. For a programme of 12 transformations and more than 150 contributors, the step no UPM source yet shows is a measured result attached to a pilot or a tool. That result would be the fair test of the order and invoice pilots in particular, since their accuracy can be counted line by line from the systems they feed.

When it doesn't apply

Annual reports and company stories describe programmes at a high level, and pilot results may exist internally without being published. Businesses with low volumes of orders and invoices may find the document-flow case weaker than it is for a group of UPM's size. The 2025 report puts that size at 15,127 employees at the end of the year, down from 15,827 at the start.

Quarri for pulp, paper and packaging is built around how a fibre operation runs, from furnish to converted order.

Quarri is an AI-native data platform for the timber supply chain. It connects buying, production, sales and inventory for forest management, sawmill, wood products and pulp, paper and packaging operators.