Altra Set a Benchmark for AI in the Public Sector
Updated: Aug 12
Artificial intelligence in public procurement may sound ambitious. At Altra, it became a practical way to shorten one of the most time-consuming parts of its work.
The organisation introduced a specialised AI agent to prepare public procurement documentation in line with Lithuania's Public Procurement Law and its internal procedures. Early testing indicates that the AI agent can reduce the time required to prepare procurement documentation by up to 80%, while maintaining the same legal and security requirements.
The work behind every procurement process
Altra is responsible for one of Europe's largest and most complex nuclear decommissioning projects. Public procurement is an essential part of this work. Long before a tender is published, teams prepare technical specifications, contract conditions and supporting documentation, all of which must comply with legislation and the organisation's internal requirements.
Preparing these documents had become one of the most demanding administrative processes within the organisation. Each document required careful drafting, review and formatting before it could proceed.
Guess what! Today, much of that work is powered by artificial intelligence.
An AI agent built around legislation
When evaluating different AI platforms, Altra was looking for a solution that could address a specific operational process. Following a recommendation from the Google Cloud team, the organisation partnered with TeraSky Europe to implement Gemini Enterprise.
The first AI agent was designed specifically for public procurement documentation. According to Mindaugas Bereiša, Head of Information Technology Development at Altra, the agent was built to operate strictly in accordance with the Public Procurement Law and the organisation's internal rules.
"Public procurement in the public sector often becomes a barrier to innovation because of its complexity and strict regulation. We wanted to change that, and we have succeeded. Now the AI agent can independently prepare and accurately format complex public procurement documentation."
Mindaugas Bereiša, Head of Information Technology Development, Altra
More time for work that requires expertise
The first results are promising. Early testing indicates that the AI agent can reduce the time required to prepare procurement documentation by up to 80%. This reduces the documentation workload for engineers and procurement specialists and helps speed up a process that is critical to project delivery and the organisation's operations.
Setting a benchmark
Altra operates in one of Lithuania's most highly regulated environments, where every new technology must meet demanding legal and security standards.
The project showed that artificial intelligence can meet those requirements while delivering measurable operational improvements, setting a benchmark for AI adoption across the public sector.
"Our collaboration with TeraSky Europe and the implementation of Gemini Enterprise demonstrated that even in a strategically important state-owned enterprise, AI can safely and efficiently handle administrative work. It is more than time saved – it changes how public procurement can be organised," points out Mindaugas Bereiša, Head of Information Technology Development at Altra.
“In the public sector, AI is often mistakenly seen as a source of security and legal risks. The Altra case shows that a properly configured AI agent can instead serve as a strong compliance safeguard. Programmed to follow public procurement laws and internal rules, the AI tool adds an extra layer of assurance and helps prevent possible human errors when preparing large volumes of documentation. The specialist then reviews and verifies the information provided by the AI.”
Greta Mieliauskaitė, Account Executive, TeraSky Europe
Altra is the strategic operational brand introduced by Lithuania's State Enterprise Ignalina Nuclear Power Plant to manage the nation's nuclear legacy. The organization is tasked with the safe decommissioning of RBMK-type reactors and the development of long-term radioactive waste repositories through 2100.
