Parallel Works adds AI governance & token budgeting

Parallel Works has added AI governance and budget management features to its ACTIVATE AI platform, targeting organisations that manage AI use across cloud and on-premises models.

The new functions let enterprises and government bodies manage access to commercial and privately hosted large language models through a single gateway. They focus on token budgeting, chargeback and usage oversight as spending on AI services rises.

Parallel Works is framing the update around a familiar challenge in large technology estates: tracking resource consumption when use is spread across departments, providers and infrastructure types. In this case, the resource is AI-model token use, which can become difficult to monitor as staff adopt multiple external and internal services.

The ACTIVATE AI governance module includes a virtual API gateway for public and private model access, real-time token usage reporting, budget allocation tools and organisation-level tracking. It also provides cost accounting and chargeback functions that can be assigned at the user, group, department or organisation level.

The platform supports OpenAI-compatible providers, Anthropic, Azure OpenAI, AWS Bedrock and privately hosted models. This gives customers a common control layer across commercial AI services and in-house systems without tying operations to a single provider.

Matthew Shaxted, Chief Executive Officer at Parallel Works, said the issue for customers is no longer only model access or model quality, but also oversight of the economics of usage.

"Organisations are discovering that the future of AI will be defined as much by governance and economics as by the model itself," said Matthew Shaxted, Chief Executive Officer, Parallel Works.

"As AI adoption expands across departments, teams and cloud providers, token consumption is quickly becoming fragmented and difficult to manage. Enterprises need centralized visibility, accountability and financial controls to ensure AI can scale sustainably across the organization."

Customer demand

The governance functions are already deployed in a large system-integrator environment operated by FutureTech, where they support thousands of users and track token consumption across a range of AI workloads. The deployment suggests demand from resellers, integrators and large public-sector or defence users that need tighter control over who is using AI tools and what those tools cost.

Chris Coker, Vice President, Major Accounts, Aerospace & Defence at FutureTech, said customers are asking for stronger controls as AI use expands into broader operational settings.

"Our customers are demanding stronger AI governance capabilities to ensure AI can be deployed securely, responsibly, and at scale," said Chris Coker, Vice President, Major Accounts, Aerospace & Defence, FutureTech.

"The combination of token budgeting, usage visibility and chargeback, integrated directly into the compute governance environment, gives our clients the controls they need to scale AI responsibly and with confidence."

Parallel Works has long focused on hybrid and multi-cloud computing management, particularly in environments with heavy compute requirements such as research, high-performance computing and government workloads. By extending that approach into AI usage tracking, the company is seeking to connect model access with the same governance structures already used for compute and storage resources.

This is increasingly important in organisations where AI tools are no longer confined to specialist teams. As use spreads across staff and business units, finance teams, IT departments and operational leaders are under pressure to understand which groups are generating costs and whether those costs should be allocated internally.

Hybrid control

Parallel Works also highlighted the integration of AI consumption governance with broader hybrid infrastructure management. ACTIVATE combines compute orchestration, GPU governance, Kubernetes management and AI usage controls in one system, giving administrators a single view across different parts of their environment.

Michael McQuade, Director of Engineering at Parallel Works, said the pace of change in AI is adding to the challenge for organisations managing hybrid estates.

"Developers consistently want the state of the art, and in AI, that's changing day by day. ACTIVATE AI gives organizations a unified governance layer across both commercial AI APIs and private infrastructure, which is critical for anyone running hybrid environments," said Michael McQuade, Director of Engineering, Parallel Works.

"Enterprises want to expand AI access across their teams, but without governance controls, costs and operational risks spiral fast."

The new governance and token-budgeting functions are now available for large enterprises, government and defence organisations, research institutions and high-performance computing environments that run private GPU infrastructure or use commercial AI APIs at scale.