The ROI Question Is Settled: 80% of Organizations Now See Measurable Returns from AI Agents

For the past two years, the conversation around AI agents has followed a familiar pattern. Every conference, every industry report, every panel discussion eventually landed on the same question: "Yes, this is impressive technology, but does it actually deliver business value?"

That question finally has a clear answer.

According to new research from Anthropic in partnership with Material, 80% of organizations now report that their investments in AI agents have generated measurable economic returns . This is not pilot, phase optimism or vendor hype. It is data from over 500 technical leaders across industries and company sizes, and it marks a definitive turning point .

The era of experimentation is over. The era of measurable ROI has begun.

The Numbers Tell a Clear Story Let us start with the headline finding because it deserves emphasis: eight out of ten organizations say AI agents are delivering real, quantifiable value . This is not confined to early adopters or tech, native companies. It spans manufacturing, retail, healthcare, financial services, and beyond.

Lenovo's global CIO Playbook 2026, conducted in partnership with IDC, reinforces this picture. The research found that organizations are projecting an average return of $2.79 for every dollar invested in AI . Nearly half of all AI proof, of, concepts have already moved into production, and 93% of surveyed leaders plan to increase AI investments in the coming year .

KPMG's Q4 AI Pulse Survey adds another layer. Business leaders expect to deploy an average of $124 million toward AI over the next year, and 67% say they will maintain spending even if a recession hits . AI is no longer viewed as experimental. It is viewed as essential.

But here is where it gets interesting. While 80% are seeing returns, the depth of those returns varies significantly based on how organizations approach deployment . The ones seeing the biggest gains are not just buying tools. They are rethinking how work gets done.

Where the Value Is Coming From

The data reveals three clear areas where AI agents are delivering measurable impact.

First, software development has become the proving ground. Nearly 90% of organizations now use AI to assist with coding, and 86% have moved beyond experimentation to deploy AI agents for production code . This is not about generating snippets. It is about agents handling real development work across the entire lifecycle.

The efficiency gains are striking. Organizations report time savings of approximately 60% across four key areas: planning and ideation, code generation, documentation, and code review and testing . When an agent handles the routine work, developers can focus on architecture, complex problem, solving, and system design.

Second, data analysis and reporting have emerged as high, value use cases. Sixty percent of organizations identify data analysis as one of their most effective applications . This makes sense. Every department produces data. Finance needs monthly reports. Sales needs pipeline analysis. Operations needs supply chain visibility. AI agents can process this work in minutes instead of days.

Third, internal process automation is delivering operational savings. Forty, eight percent of organizations report significant value from using agents to automate repetitive workflows . These are the tasks that slow teams down without requiring deep expertise, data entry, form processing, status updates, handoffs between systems. When agents handle these, human workers can focus on work that actually requires judgment.

The Split Between Leaders and Everyone Else

Here is where the data gets more nuanced. While 80% see returns, not everyone is seeing the same magnitude of returns.

TheCUBE Research found that organizations with proper governance and frameworks in place are achieving an average ROI of $3.70 per dollar invested . That is significantly higher than the $2.79 average reported in broader surveys. The difference comes down to readiness.

KPMG's research shows that nearly two, thirds of leaders (65%) cite agentic system complexity as their top barrier . Organizations that invest early in governance, data quality, and integration capabilities are pulling ahead. Those treating AI as an add, on are getting stuck in pilot purgatory.

Deloitte's Finance Trends 2026 survey adds a cautionary note. Among finance leaders actively using AI, only 21% believe those investments have already delivered clear, measurable value . This does not contradict the broader 80% data but it reflects that different functions are at different stages. Finance, with its sensitivity around data accuracy and compliance, is moving more carefully. But even here, the trajectory is clear. The same survey found that 87% of CFOs believe AI will be extremely or very important to their finance operations this year .

What Success Looks Like in Practice

Apart from the surveys, tangible examples illustrate what measurable ROI really looks like.

Algar Telecom, a Brazilian telecom company is another case in point. In 2024, they launched an AI agent “Billy” to review first invoices for new customers – a notorious pain point where the vast majority of billing errors happen. Over the first nine months of 2025 Billy reviewed 25% of all initial invoices and added an extra $1.5 million to earnings. That is measurable ROI.

Or consider Thomson Reuters, which uses AI agents to power its legal platform CoCounsel. Tasks that once required lawyers to spend hours manually searching documents now take minutes. The agent can access 150 years of case law and 3,000 domain expert resources instantly .

In cybersecurity, eSentire reduced expert threat analysis time from five hours to seven minutes. Their AI, driven analysis now aligns with senior security experts 95% of the time .

In retail, L'Oréal achieved 99.9% accuracy in conversational analytics, with 44,000 monthly users able to query data directly instead of waiting for custom dashboards .

These are not pilot projects. These are production systems delivering real financial outcomes.

The Skepticism Is Still There, And That Is Healthy

Not every source agrees that the ROI case is settled. Jefferies, citing surveys from Google Cloud and Anthropic, notes that while agent usage is up 50%, many companies still struggle to prove clear returns . Reliability issues remain, agents can fail, loop endlessly, or take wrong actions. Inference costs add up when running models at scale.

This is a healthy tension. The 80% figure comes from organizations that have already deployed agents and measured results. The skepticism comes from those still evaluating or running limited pilots. Both perspectives are valid.

The truth is that ROI is not automatic. It depends on use case selection, implementation quality, and organizational readiness. As KPMG notes, the topline adoption number can undersell what is actually happening among leaders . The ones pulling ahead are not just deploying agents. They are professionalizing their approach, building governance, investing in infrastructure, and preparing to run multi, agent systems reliably.

What This Means for 2026 and Beyond

AI agents have crossed the chasm from experimental technology to core business infrastructure.

For organizations still on the sidelines, the risk is no longer "will this work?" It is falling behind competitors who are already compounding their returns. Lenovo's research shows that while 60% of organizations are in late, stage AI adoption, only 27% have comprehensive governance frameworks in place . That gap represents both a warning and an opportunity.

For organizations already seeing returns, the focus shifts to scaling. KPMG found that 81% of organizations plan to explore more complex agent applications in 2026, including cross, functional workflows and multi, step processes . The companies that figure out how to orchestrate multiple agents across different functions will build advantages that are difficult to replicate.

The question is no longer whether AI agents deliver value. The data says they do. The question is whether your organization is ready to capture that value at scale.