Anthropic adds spend-tracking tools as enterprise AI buyers demand proof of return

Anthropic said Wednesday it is rolling out new enterprise reporting tools meant to help customers show what their AI token spend is actually producing. The move is aimed at chief financial officers, engineering leaders and other buyers who want clearer evidence that their budget for Claude-based workflows is translating into output, efficiency and business value.

The new suite lets managers review team output, costs and workflow breakdowns over a chosen date range, using internal analytics and employee survey data. Anthropic is also adding pooled group budgets for token management, a sign that the company wants to make AI consumption easier to monitor and control inside larger organisations.

The product push comes as Anthropic tries to deepen its hold on enterprise customers even as competition from OpenAI remains intense. It also lands at a moment when the company is showing stronger financial momentum, including investor-facing results this week that TechCrunch said showed operating profit in the second quarter.

What Anthropic launched

Anthropic’s new reporting tools are designed to make AI usage legible to business buyers who are under pressure to justify spend. Instead of treating token consumption as a black box, the company is packaging data in a way that should help managers assess what they are getting back from model use.

According to TechCrunch, the reporting suite can surface team output, the cost of that output and the workflow steps behind it. The reports draw on Anthropic’s internal analytics as well as employee surveys, which suggests the company is trying to capture both quantitative usage data and the human side of adoption.

The company is also bundling recommendations intended to improve results. In practical terms, that means the software is not only reporting on performance but also nudging customers toward more efficient or more productive use of Claude-based tools.

How does this help enterprise buyers?

It gives them a clearer line of sight into return on investment. The core problem for many large buyers is not whether teams are using AI, but whether those tools are delivering measurable savings, faster throughput or better-quality work.

That question matters because token spend can ramp quickly once an organisation standardises a model across departments. Anthropic’s reporting tools appear designed to answer the questions that finance teams ask after adoption has already spread: who is using the tool, what is it costing and what is it producing?

Why it matters

The new tools show how AI vendors are moving beyond raw model access and into the management layer around usage, cost control and retention. For enterprise buyers, that is significant because the next phase of AI adoption is less about experimentation and more about proving value at scale.

Anthropic is clearly aiming at that audience. The company’s pitch is not just that Claude can help teams work faster, but that it can be measured in a way CFOs and engineering leaders can live with. In a market where many organisations are scrutinising software budgets more closely, that kind of visibility can matter as much as model quality.

The move may also strengthen customer lock-in. Once teams build workflows around a model or tool, switching costs rise. That is especially true in developer environments, where standardised patterns and embedded processes can make replacement expensive and disruptive.

For Anthropic, this is also a retention play. By giving customers reporting, guidance and budget controls, the company is trying to make itself harder to remove from enterprise stacks. That matters because enterprise usage is where recurring revenue, expansion and long-term contract value are most likely to build.

By the numbers

Background: how Anthropic got here

Anthropic’s latest product decision fits a broader shift in enterprise AI buying. Token spend is now a common cost measure for large language model services, and that makes it possible for finance teams to compare usage against budgets, headcount and output.

That same logic has made AI coding tools especially important. Once a development team embeds a tool into day-to-day work, the vendor is no longer selling a novelty; it is selling a layer of the workflow. Anthropic’s Claude Code appears to have become one of the clearest examples of that dynamic.

TechCrunch reported that Claude Code is a major contributor to Anthropic’s business momentum. The company has also shared investor-facing financials this week showing operating profit in Q2, reinforcing the idea that enterprise demand is translating into real business performance.

Jellyfish’s report, published Tuesday, offers a useful snapshot of that demand. The software company said Claude Code and related Anthropic-model tools were used for nearly one-third of proposed code changes among its customers, and that spending on those tools is up 10x since the start of the year. Jellyfish estimated the tools account for about 2% of total engineering cost.

That balance is important. Two percent of engineering spend may sound manageable on paper, but the jump in usage can still catch finance teams off guard if they do not have clear reporting. That is exactly the gap Anthropic is now trying to fill.

What the Jellyfish data suggests

Jellyfish’s findings help explain why Anthropic would invest in reporting features now. The data point to a product category that is spreading quickly inside engineering organisations, but not always with equally quick budgeting discipline.

According to TechCrunch, Nicholas Arcolano, Jellyfish’s chief scientist, said:

“Even if you’re saying 2% is still small, that 10x growth is growth that most CFOs didn’t plan for.”

That statement captures the tension behind Anthropic’s launch. Rapid adoption can be a success signal for the vendor and a cost-management headache for the customer. Reporting tools are the bridge between those two realities.

What the new tools are designed to do

Anthropic’s reporting suite is built to help teams answer the questions that determine whether AI stays in the budget. The tools let managers inspect output, costs and workflows over a date range they choose, which could help them compare teams, projects or periods of adoption.

The company is also adding pooled group budgets for token management. That suggests Anthropic wants enterprise administrators to be able to set shared limits rather than leaving spending to individual teams or users.

The reports reportedly include recommendations for improvement as well. That is a subtle but important shift: Anthropic is not only providing accounting-style visibility, but also positioning itself as an advisor on how to use its models more effectively.

TechCrunch said the new tools resemble lighter-weight forward-deployed engineering support. In other words, Anthropic is packaging some of the hand-holding that large enterprise customers often get from vendor teams into a product layer that can scale more easily.

Enterprise demand is changing the product playbook

Large AI buyers are increasingly asking vendors to prove value, not just provide capability. That is true across software procurement, but it becomes even more pressing when usage-based pricing means costs can rise in direct proportion to activity.

Anthropic’s move reflects that pressure. Instead of relying solely on model performance or brand perception, the company is making a case that it can help customers measure usage in business terms. That is a different kind of sales pitch, and one that may resonate with larger organisations trying to reconcile innovation budgets with finance controls.

It also suggests that the competitive fight with OpenAI is not only about model capability. It is about which vendor can build the strongest enterprise operating system around those models — dashboards, budgets, recommendations and the support tools that make adoption sustainable.

TechCrunch reported that Anthropic remains ahead with enterprise customers despite competition from OpenAI models. That advantage may rest partly on the kind of buyer trust that comes from showing not just what an AI can do, but what it costs and what it returns.

Why switching costs could rise

Once an organisation standardises a model or tool across teams, replacing it becomes harder. Developer habits, internal workflows and reporting systems all become tied to the vendor’s ecosystem.

Anthropic’s new tools may deepen that effect. If customers begin relying on its reporting, budget management and workflow recommendations, then Anthropic is no longer just a model provider. It becomes part of the machinery used to manage AI spend and measure performance.

That makes the product strategically important. The more a vendor can sit inside the customer’s operating rhythm, the harder it becomes for a rival to pull that customer away with a better price or a slightly better model.

Jellyfish AI product lead Nikolas Albarran put the enterprise logic bluntly, saying:

“Anthropic was just there with great enterprise motions”

That line helps explain why Anthropic’s latest launch is less about a flashy new model feature and more about the infrastructure around enterprise adoption.

Timeline

What happens next

The biggest open question is whether the new tools will materially change how customers manage usage or simply make spending easier to observe. Anthropic has not disclosed how much additional revenue it expects from the reporting suite, and it is not clear whether customers will use the new visibility to cut back or to expand adoption more confidently.

Another question is whether rivals will respond. If enterprise buyers begin expecting token-spend reporting, workflow analytics and budget controls as standard features, OpenAI and other vendors may need to match those capabilities more directly.

It will also be worth watching how broadly Anthropic rolls the tools out across its enterprise base. The larger the adoption, the more likely the reporting layer becomes a meaningful part of the company’s retention strategy.

For now, Anthropic is betting that the future of enterprise AI sales depends on clarity as much as capability. In a market where usage can grow quickly and budgets can lag behind, making AI spend legible may be just as valuable as improving the model itself.

Key questions readers may still have

How does Anthropic’s reporting suite work?

It combines internal analytics and employee survey data to show managers team output, costs and workflow breakdowns over a selected date range, while also offering recommendations to improve results.

Why is this launch important for enterprise AI?

It addresses a major buyer concern: whether AI spend is producing measurable value. That matters for CFOs, engineering leaders and procurement teams trying to justify usage-based costs.

What makes Claude Code strategically important?

TechCrunch said it is a major contributor to Anthropic’s business momentum, and Jellyfish’s report suggests it is already embedded in developer workflows at scale.

Could this affect competition with OpenAI?

Yes. If buyers value spend visibility and workflow management as much as model performance, enterprise tooling could become a larger part of competition between vendors.

Anthropic’s latest product update is therefore about more than reporting. It is an attempt to turn AI spend data into a retention engine, and to make the economics of enterprise adoption easier for customers to defend internally.