Harness Launches Two Products to Give Business Teams Full Visibility into ROI of AI Spend

Almost all engineering leaders surveyed – 94% – did State of Engineering Excellence Report they say they don’t get cost metrics from their current measurement frameworks, making it difficult to be sure that every dollar spent on AI is producing a real result.

According to Harness, there are two problems driving the use of AI: developers use it to generate almost all of their new code, and infrastructure agents spend large amounts of tokens on ticket resolution, customer interaction and workflow automation, without any real sense of how much value is being gained from that spending.

“There’s a lot of talk right now about the rising costs of AI. But the real challenge isn’t the spending itself—it’s that teams can’t put down the cost impact,” Trevor Stuart, senior VP at Harness, told SD Times in a statement.

So Harness is releasing two new products that help organizations get a handle on how much money is being spent on AI, and whether or not companies see the value of that. The first is AL DLC Insights, and the second is Cloud & AI Cost Management.

AI DLC Insights

According to the company, AI DLC Insights allows organizations to see where spending on tokens is resulting in outsourced work, and where it isn’t. An agent running in the developer’s IDE captures each line of code, records the token cost and maps it back to a pull request or ticket or submitted task. With that, companies can see, for example, that it costs X amount in AI credits to fix a bug, and they can decide if that’s more expensive than having humans do that work.

“Companies pay to write code that never reaches end users. That’s the problem AI DLC Insights solves,” explains Stuart. “It’s an agent embedded in developers’ machines that captures tokens at source, uncovers performance gaps, and delivers informed recommendations. You get a clear view of ROI, efficiency, and actual cost per feature or bug resolved.”

The following abilities are included in this AI DLC Insights release, Harness announced:

• Unified AI code adoption visibility – One place to track adoptions, timelines, and AI-generated code across code agents – Claude Code, Cursor, GitHub Copilot, Windsurf. What tools are your engineers using, not just what chairs you bought.
• Per Developer Attribution – Token usage, times, and code sent are tracked across developers, agents, warehouses, teams, and business units behind you, turning multiple AI invoices into per-developer ROI.
• Wasted spend recovery – Tokens burned in scrapped code, bloated commands, expensive model selections, and missed cache hits appear automatically. The first time a team doubles its token charge without sending more code, you know before the next update.
• Code-to-production impact — Track AI-generated code from rapid to production using ship rate, PR cycle time, and DORA metrics, correlated with incident and vulnerability data. Know that coding agents actually make your team faster.
• Measurement and governance – Adoption, efficiency, and impact metrics compared across teams against an organizational baseline, with role-based access control and native engineering management.

Cloud & AI Cost Management

The second feature, Cloud & AI Cost Management, picks up where development cost visibility ends. In its announcement of the new products, Harness wrote: “The monthly spend of $28,000 on a customer support agent is a completely different number depending on how many tickets they solve. If it costs $0.60 per ticket solved and the other human exception is more expensive, it is one of the best investments in your stack. If the math works out the other way, you’re paying more for the difference for an organization that can’t tell an organization automatically.

In its announcement, the following capabilities are included in this release of Cloud & AI Cost Management:

• Integrated AI cost visibility – One place to see spending across all AI and managed service providers, from OpenAI and Anthropic to AWS Bedrock and GCP Vertex AI. The single source of truth for every AI dollar spent, regardless of where it comes from.
• Full revenue attribution – Costs are tracked down to the agent, session, workflow, team, and operating business unit, converting total invoices into actionable agent ROI.
• Anomalous detection – Use the spikes marked before they reach the invoice, using the same detection engine already watching your cloud costs. The first time the release doubles your bill of tokens, you know before the finance does.
• Budgeting and management — Budgets are set at the agent, team, or business unit level, with approved model policies and Cost Categories that extend the same FinOps controls you already trust with cloud AI.

“A lot of what we build at Harness starts with our inner pain,” says Stuart. “With both products, we were hearing the same thing from customers and watching the industry struggle with the rising costs of AI – but we also needed to solve it ourselves.”