Geordie launches Cost Intelligence for AI spending

Geordie has launched Cost Intelligence, a product that links AI spending to the agents and workflows behind it.

The move extends Geordie's work in agent oversight into financial monitoring as businesses seek clearer explanations for rising AI bills.

AI spending is often tracked through model usage, token volumes and invoices, but those measures do not show which tasks or decisions drove the costs. Cost Intelligence is designed to help AI operations teams trace spending back to individual agents, the activities they carried out and the workflows they supported.

The issue is becoming more acute as AI tools spread across organisations and costs vary from one agent run to another. In one case cited by Geordie, a retailer lost USD $1 million a month through model-routing waste when tasks were sent to more expensive models than necessary. In another, a bank used 50% of its monthly AI budget in a single day after an agent became stuck in a tool loop.

One customer said the product changed how it assessed AI spending.

"One of our biggest challenges with AI spend has been visibility," said Jacob Sweat, VP of Technology, Wood Partners.

"Once Geordie gave us a clear view of what agents are costing us and whether that spend is doing something useful, the conversation shifted from 'how much are we spending?' to 'is this spend worth it?'" Sweat said.

Broader risk

Geordie has so far focused on security and governance around AI agents. Its system sits close to the agent itself, allowing it to observe behaviour in detail and apply that same view to spending, token use and model selection.

That means teams investigating a jump in costs can identify which agent acted, what work it performed and who was responsible. This gives finance, operations and security teams a shared record of how AI systems are behaving inside the business.

Henry Comfort, Co-Founder and Chief Executive Officer of Geordie, said existing AI cost metrics stop at consumption.

"Tokens and invoices tell you what you consumed; they do not tell you what work drove that consumption," said Henry Comfort, Co-Founder and Chief Executive Officer, Geordie.

"Cost Intelligence connects spend back to individual agents and their activity. When you can see what an agent costs alongside the work it performed, you have the foundation for understanding agentic ROI, and that is exactly what Cost Intelligence delivers," Comfort said.

Detailed view

The product aggregates spending across connected platforms and attributes it to the work that generated it. Users can move from an organisation-wide total to narrower views by platform, team, user, model, agent or workflow.

It also tracks cache efficiency at the level of the individual agent, which can show where poor caching is increasing costs. Another feature maps spending and usage volume by agent to distinguish between agents that are expensive because they are busy and useful, and agents that are generating waste.

The system can also flag cost anomalies such as infinite loops, oversized model selection and sessions that consume large volumes of tokens without producing useful output.

These controls matter because AI agent risk is widening beyond cybersecurity. As organisations allow agents to handle more tasks with less human intervention, oversight is shifting toward operational and financial exposure as well as technical safety.

"As agents take on more autonomous work across the enterprise, the risk they create isn't limited to security alone; it's operational, financial, and accountability risk," said Hanah-Marie Darley, Co-Founder and Chief AI Officer, Geordie.

"Organisations need a shared view of how agents are operating and the full range of risk their behavior creates, not just the security dimension," Darley said.

Cost Intelligence is generally available, giving customers a way to examine AI spending across cloud, code and endpoint environments rather than only where activity passes through a single gateway or model router.