AI has entered the “prove it” stage. Nearly a third of all companies invested at least 1.7% of revenue in AI last year, according to BCG research. In 2026, that figure is closer to two-thirds. But the investment and enthusiasm behind AI is outpacing measurable returns. Consider that 60% of companies report minimal or no value—including cost reductions or revenue gains—despite significant efforts.
Meanwhile, risks and expenses are increasing, with nearly two-thirds of companies reporting uncontrollable AI scaling expenses and a similar share encountering hallucination and explainability challenges. Roughly three-fourths point to security concerns and challenges from unstructured data.
Companies are spending without realizing expected cost gains.
But some companies are pulling ahead. These leaders recognize that their AI and cost reduction agendas are inseparable. By tightly linking AI deployment to structural cost transformation, they are compounding advantage and building operating models that will widen the competitive gap as AI capabilities accelerate.
Another BCG analysis reached similar conclusions: AI leaders deliver 3 times greater cost reduction, 1.6 times higher EBIT margins, and 2.7 times the return on invested capital than their peers. Beyond cost advantages, leading companies are increasing transparency across value chains, getting faster at decision making, and unlocking capital to reinvest in growth and innovation. These companies show what’s possible, and they have insights for how others can catch up.
Where Companies Go Wrong
AI-driven cost reduction isn’t just a tech problem. It requires rethinking processes, operating models, and how work actually gets done. Even without AI, cost programs often fall short because companies take reactive, short-term actions that don’t fundamentally redesign workflows, operating models, or culture. Although these programs may succeed in the short term, costs inevitably creep back. And far from guaranteeing success, deploying AI brings new complexities and can even cause costs to rise.
We see a common set of challenges across cost programs:
Too many fragmented initiatives, not enough scale. Many organizations are experimenting with AI and launching proofs of concept across the entire company. That enthusiasm is understandable, but without clear priorities, companies risk diluting their efforts, increasing overhead, and applying AI to areas where it might not generate the biggest impact.
Foundational issues with data and technology. Even when AI pilots succeed, they can be challenging to scale across the enterprise because of technology and data limitations. Testing and resiliency requirements for a company-wide implementation are far more complex than for an isolated initiative. Organizations that don’t have a solid foundation in place—in terms of their data and tech solutions and architecture—must compensate by keeping many people in the loop to check outcomes, eroding value.
Insufficient focus on training and upskilling talent. Some companies implement a new AI initiative only to see employees ignore it, often because they lack the skills and capabilities required to use the new tools. Even the companies that launch training programs sometimes fail to tie AI to real workflows and codify new ways of working and expectations.
Failure to redesign workflows and processes. All too often, companies settle for incremental efficiency improvements from their AI. That means they miss out on the kind of transformative gains that the technology can deliver. Time and time again, we find that in a typical AI implementation, only 10% of the value comes from the algorithms and 20% comes from the technology and data. The remaining 70% comes from managing process change—mainly from redesigning workstreams and processes end-to-end. That’s a more complex endeavor, but it delivers a much bigger cost impact.
A lack of clarity on how to turn efficiency gains into financial value. Even when organizations can point to specific efficiency gains from AI, such gains often don’t impact P&L. Without a clear plan and tracking infrastructure, efficiency gains vaporize and don’t translate into a cost advantage.
Stay ahead with BCG insights on cost management
The Roadmap for Success
Leaders are focused on how to launch AI initiatives while also using AI as part of a deliberate sequence of traditional cost levers. Their goal is to deliver results immediately—and, ultimately, systematically. This requires striking the right balance between generating near-term momentum and deliberately orchestrating initiatives toward a cohesive, scaled ambition.
Based on BCG’s growing body of work in helping companies implement AI-enabled cost transformations, we’ve identified four key success factors to generate better results.
Start with proven deployments to fund the journey. Rather than trying to implement the technology hastily across the entire enterprise, companies should start with a small number of workflows based on relatively mature solutions that yield rapid results. This approach generates momentum, builds AI capabilities, and unlocks financial gains that can be used to fund longer-term measures.
Procurement is a good example. Companies spend significant amounts on procurement, the transactions are relatively straightforward, and the range of potential issues is small. For those reasons, companies can in many cases use commercially available AI solutions to rapidly improve performance. AI can compile and synthesize purchasing information from the entire organization, identify outliers, determine where the company is overpaying or underperforming in terms of working capital, and propose solutions for buyers—all while improving visibility.
Tasks that were once handled through manual reviews and fragmented data analysis are increasingly being automated and enhanced with AI. Specific applications and outcomes that we’ve seen include:
- Supplier reviews to optimize the base of suppliers, standardize pricing, and negotiate for discounts: 5% to 25% savings in three to six months.
- Specification review to reduce the specs for products and services: 5% to 10% savings in three to six months.
- Inventory optimization: 5% to 15% savings in three to nine months.
Other areas where AI applications can yield immediate results include marketing analytics and processes, software engineering, field support for sales teams, customer service centers, product development, and finance processes.
Reinvent workflows and processes for greater impact. While companies can generate fast, incremental results by applying AI to their existing processes, the true value comes from reinventing workflows. The goal is to integrate processes across different departments and functions, capitalizing on digital and AI to dramatically increase efficiency.
This is a bigger endeavor, and one where companies sometimes underestimate the difficulty. But it can lead to breakthrough gains in productivity, efficiency, and value creation. Reinventing processes can generate three to four times the impact of traditional incremental improvements.
Because this redesign process requires collaboration across functions, it can be complex. Companies can increase their odds by picking one process and designing it from scratch, end-to-end, across the entire value chain.
Apply agentic AI in the right situations. AI agents are systems that can observe, plan, and act to achieve defined goals and learn from continuous feedback. The technology holds massive potential, because agents can act autonomously rather than just provide insights—which could enable major cost reductions, especially in functions like HR, finance, customer service, and IT.
Despite this promise, some companies have yet to capitalize on agentic AI. In our experience, the usual limiting factor isn’t the technology but the choice of applications. Teams try to stitch it into existing processes, with constraints on legacy tech stacks that have inaccessible data. In fact, the sweet spot for agentic AI applications is in complex processes and environments where risk exposure and ethical or governance sensitivity are comparatively low. In less complex situations, baseline automation technology solutions are sufficient. In areas with stringent risk and compliance requirements, human oversight is a must.
For example, BCG worked with a global consumer goods company to develop a custom AI agent that could reshape marketing and development processes. The goal was to identify promising insights and move them from concepts to in-store offerings quickly. Ten custom agent workflows were rolled out to more than 500 users across the company’s global operations—all with human-in-loop controls and safety guardrails.
Overall, the project reduced the time spent on key workflows by 25% to 40%, helped get new products and marketing executions into the market twice as fast, and generated substantial efficiency gains in the P&L. The company reports that user satisfaction with the agentic AI solutions is higher than 90%.
Other agentic AI examples include:
- Design. A shipbuilder used agents to accelerate a core engineering and design task. The work itself was complex and bespoke, requiring the coordination of multiple variables and calculations. To address this, the company deployed a multiagent AI setup—combining a conversational interface with specialized engineering agents—to support the full workflow from technical specifications to design outputs. Engineers remained in the loop to validate results, but the system significantly accelerated what had traditionally been a time-intensive, manual process. With agentic AI, the company reduced lead times for design elements from five days to just one day and cut engineering costs by 45%, while producing outputs that were as accurate (or more) than traditional approaches.
- Software modernization. A bank in Asia-Pacific used agents to analyze the code that supports its response to user queries and to provide retro-documentation of the process. The tool reduced the time it took engineers to understand the existing code by up to 30%, with time savings projected at more than 70% when rolled out.
- Supply chain scenario planning. An industrial goods manufacturer used AI agents to get better visibility over its complex global supply chain and help managers make more objective, data-driven decisions. The tool compiles real-time information across all inputs and production sites and assesses the impact of different scenarios on cost, production time, and other variables.
- Customer service. An insurance company wanted to improve customer-support response times, but it was limited by internal inefficiencies. (For example, employees often had to retrieve information for more than ten systems that weren’t linked to each other. Training them how to do that took months.) To improve, the company developed an agentic AI pilot that could pull that information from disparate sources. Employees were 35% more efficient in answering customer queries after just 45 minutes of training, and junior employees were much more likely to be able to field complex questions on their own, without having to redirect them to more senior people.
Rigorously track value. Many companies are seeing efficiency gains from AI pilots, but that’s only the first step. To generate meaningful value from the technology, leaders need to link those improvements to bottom-line impact in the P&L. That entails building a clear business plan with specific metrics, timelines, and projected ROI. Moreover, teams need to make strategic decisions for how freed-up staff time can be reallocated.
For example, if AI makes a specific activity 15% more efficient, teams handling that activity could become 15% smaller, or increase their capacity to take on 15% more work in other value-creating areas. Managers may even decide to give employees that time back to improve morale. Regardless of how it’s handled, senior leadership teams need to think through these implications.
Any initiative should be thought through in terms of its full P&L and balance-sheet impact before implementation—which specific line items are expected to move, up or down, and over what time horizon. Companies should have a clear sense of which line items will increase or decrease, along with timelines for that impact. Then, once the initiative is in place, companies track what is happening from the bottom up (“Is the value materializing in our financials?”) and top down (“Do I expect other impacts soon?”).
More broadly, metrics can and should go beyond sheer task efficiency to directly align with business outcomes, such as reduced cost-to-serve or working capital improvements.
How Companies Are Using AI to Drive Results
Leading companies are already translating AI ambition into tangible results across core functions. Their experiences show how targeted deployments can unlock efficiency, reduce costs, and fund broader transformation.
Reinventing Marketing with GenAI. A consumer goods giant built a suite of GenAI tools to improve marketing. The solutions cover tasks ranging from media and insights to innovation, content creation, and brand-performance reporting in real time. They were rolled out on a new technology platform and automated tasks that accounted for 30% to 40% of time for marketing employees. As a result, the time spent on routine activities dropped by up to 90%, while the quality of outputs doubled. Based on that early success, the company is now expanding the GenAI program to its R&D function.
Using AI to Reduce Warranty Claim Costs. Car manufacturers and parts distributors are increasingly struggling to manage warranty processes. Inflation, a shortage of technicians, and a growing number of claim anomalies have all driven up customer warranty costs. Aging legacy systems often can’t effectively identify claim risks and other issues.
In response, companies are deploying AI-based solutions that seamlessly integrate into existing warranty processes without requiring major redesign. Players are finding that claims management is emerging as a promising AI application. Companies have dynamic data inputs to manage, decisions require significant expertise, human performance and decision-making can vary widely, and processes are labor-intensive and inefficient.
At a high level, the AI system triages incoming warranty claims, validating data across customers, vehicles, dealers, repair shops, and financial metrics. The goal is to determine which claims are straightforward and can be automatically approved, versus those that require more scrutiny and should be flagged for a manual review.
Using this tool, a handful of OEMs and auto distributors are reducing warranty claim costs by about 6.5%, enough to pay off the initial project investment in three to four months.
Leading a Multibillion-Dollar AI Cost Transformation. IBM took a comprehensive approach to AI and has since reduced its annual operating costs by more than $4.5 billion. The overall goals of the transformation have been to eliminate unnecessary work, simplify processes, and automate through technology. The effort has focused on three main cost centers:
- General and administrative processes. IBM rightsized the HR function to global benchmarks, ensuring that it still had the right resources to support employees and teams. More than 90% of HR inquiries are now resolved through an AI-enabled chatbot. Operating expenses in HR dropped by 40%, even as a customer loyalty score improved by 74 points. The company also redesigned the quote-to-cash process to capitalize on AI and reduce the workload on employees. Costs for financial planning and analysis dropped by 35%. And leaders developed proofs-of-concept to use GenAI in legal, procurement, and other functions.
- IT applications and infrastructure. In addition, the company shifted to strategic platforms rather than isolated solutions. It used AI to create more visibility over total applications and infrastructure labor across the company and reduced headcount as a result. Total annual cost reductions in IT alone were roughly $600 million.
- Third-party spending. In software, AI helped the company consolidate spending by eliminating long-tail solutions and unused licenses. Leaders reduced the number of vendors and contractors, shifted some to offshore, lower-cost locations, and reduced the hourly rates for others, based on a benchmarking analysis.
Since the company announced the cost effort, free cash flow has more than doubled, and the stock price has since increased dramatically.
The most successful companies treat AI and cost transformation as a single, integrated strategy. AI holds tremendous potential to accelerate efficiency and reshape how work gets done, but capturing that value requires time to redesign workflows, modernize technology, and build new capabilities across the organization.
That journey must be funded. Traditional cost transformation provides the means to do so, generating the resources and discipline needed to scale AI effectively. When sequenced together—using early savings to finance deeper reinvention—companies can sustain momentum, capture lasting cost impact, and build a durable competitive advantage.
