As Foundational AI Improves, Information Services Providers Can Thrive

AI makers are improving their generative AI (GenAI) tools rapidly. Leveraging significant advances in large language models (LLMs) as well as users’ growing familiarity with such tools, these AI companies are tailoring their general-purpose technology to meet the needs of professional services firms.

A year ago, BCG’s survey of GenAI tool usage in professional services organizations found that tools designed specifically for knowledge workers produced much higher-quality output than general AI tools. This year’s survey, however, turns the table—with general-purpose AI tools producing output on par with specialized tools on many dimensions. (See “About the Survey.”)

This shift poses a challenge for information services providers, which supply specialized data, research, and analytics to legal, financial, tax and accounting, and other professional users. The growing capabilities of LLMs and other foundational AI models are likely to create winners and losers among information firms. Those that continue to invest in proprietary data and insights—and meet customer expectations for AI-enabled delivery—will remain central to professional services organizations. Firms that do not offer differentiated content or fail to innovate in business and delivery models risk being left behind.

The survey findings also have important implications for the professional services industry. (See “AI-Focused Actions for Professional Services Firms.”)

Big AI Players Raise the Stakes for Information Services

Information services providers have long been valued by customers and investors for their mission-critical data, deep integration into workflows, and keen understanding of industry-specific needs. Drawing on these strengths, most providers embraced the AI opportunity, launching some 200 new specialized tools and features over the past few years. A host of AI-first players also joined the landscape, introducing sector-specific solutions such as tools for law or investment banking.

Initially, specialized AI tools had the edge in professional services. In 2024, these tools outperformed general-purpose tools on typical knowledge work tasks by 24 percentage points as measured by the amount of human rework required. Over the last year, makers of general AI tools started closing this gap. (See Exhibit 1.) In 2025, general-purpose tools performed 9 percentage points higher on “no rework” output. Moreover, they scored neck-in-neck with specialized tools on dimensions such as overall satisfaction, accuracy of output, and completeness of insight. While the results were broadly consistent across user segments, specialized tools fared comparatively stronger among legal users than in finance or tax and accounting.

Does this spell the end for information services’ advantage in AI tools for these markets? Not necessarily. As foundational AI improves, the sources of advantage for information services providers come into sharper focus. Several durable advantages remain:

Data quality matters more than ever.
Professionals are increasingly aware of the importance of high-quality data inputs. In a world where “AI slop” is ubiquitous, information services firms continue to be known for producing complete, accurate, and credible data and insights, and users want to make sure their AI tools are drawing from quality sources. A significant portion of survey respondents cited the need for higher-quality data inputs as a top barrier to greater use of AI in their work: 28% in 2025, versus 19% in 2024.

Specialized tools excel for use cases that rely on domain-specific knowledge.
Generalized AI tools tend to be used for general tasks, such as qualitative analysis (which includes basic search and summarization) and writing emails. In contrast, specialized tools are used where content and context matter, such as building presentations and consulting internal documents. As knowledge workers turn to AI tools for increasingly complex use cases, with a higher bar for accuracy, they may find that specialized tools are a better fit.

AI features in specialized tools improve satisfaction and productivity.
Across professional services, employees who leverage the AI features in specialized tools have much higher customer advocacy scores than those who do not. (See Exhibit 2.) This finding suggests that the AI features integrated into information services offerings improve satisfaction and allow users to get more out of the products. Providers’ continued investment in specialized AI products and features will be critical for user activation and retention—and for long-term relevance.

Adoption and Perceived Value of AI Are Increasing

The survey also revealed some important findings about how usage of general-purpose and specialized AI tools has evolved over the last year.

Adoption rates.
Deployment of general-purpose tools in the professional workplace—such as OpenAI’s ChatGPT, Microsoft Copilot, Claude by Anthropic, and Google’s portfolio of AI solutions—is clearly growing. While ChatGPT remains the most widely used and liked, people are now branching out and leveraging multiple tools. In 2025, professionals regularly used an average of three different tools, versus just over two in 2024. (See Exhibit 3.)

Use of specialized GenAI tools has also expanded significantly. According to the survey, it rose from just over one-third of respondents in 2024 to almost two-thirds in 2025. We believe that usage could be even higher, since the data likely undercounts individuals who did not realize they were using AI features embedded within familiar tools.

More-complex tasks.
Respondents are also using both general and specialized AI tools for tasks that are inherently more complex, such as quantitative analysis, compared to basic tasks like drafting emails that predominated in the 2024 survey.

Time savings.
In both the 2024 and 2025 surveys, respondents said that the main benefits of GenAI are “increased productivity” and the “ability to do more work.” One would expect that growing use of AI would be linked to greater time savings. Yet reported time savings increased only modestly over the past year. (See Exhibit 4.)

While this slight uptick reflects progress, it likely understates the long-term potential for AI tools to help workers save time. In our survey, “adoption” ranges from using GenAI multiple times a day to as little as once a month. As a result, even though respondents report increased usage, many may not yet be deploying these tools consistently enough, or deeply enough within workflows, to realize substantial time savings.

Adoption should therefore be viewed as a leading indicator of future productivity gains. As tools continue to improve and users become more adept at integrating them into their work, time savings are likely to increase. This belief is shared by 40% of respondents, who expect time savings of 50% or more in just two years.

Charting the GenAI Future of Information Services Firms

To thrive in the years ahead, information services providers will need to leverage their specialized focus and long history of customer relationships. Five actions will be key:

Be a first mover in developing compelling AI-enabled features.
The providers that were first to launch GenAI features like semantic search and chat-with-content gained publicity and recognition—while creating a marked improvement in the user experience. This is reflected in customer advocacy scores. Those that are just now launching similar features are playing catch-up. But as the frontier of innovation continues to move, there will be opportunities to delight customers with the next generation of features.

Pursue a dual path to distribution.
In addition to deepening direct customer relationships, information services providers are creating partnerships with AI companies that give LLMs access to the providers’ high-quality data. This setup allows customers to get specialized responses through ChatGPT or another general tool, creating a second distribution channel for providers.

Some providers are already pursuing this strategy: At least five partnerships with AI companies were announced in 2025 Q3 and Q4 earnings calls, including with OpenAI, Google, and Anthropic. More such alliances are likely on the way.

In setting up these partnerships, providers need to strike a balance between making data more accessible to users and protecting intellectual property. They need to be present in the platforms where work is increasingly taking place, while ensuring that their brand stays visible to end users. This can be achieved through “bring your own license” models, where access is granted only to subscribers, or participating in AI companies’ marketplaces.

Revamp the business model.
The threat from AI is not that information services data and insights will become obsolete, but that there may be fewer human users to interact with them. For providers with seat-based models, revenues may decline as a result. These firms should consider developing machine-to-machine or outcome-based business models that separate compensation from the number of human users.

Use AI to increase the speed and volume of content creation at lower cost.
For an industry with already high margins, the case for using AI in internal operations is about producing more of what customers want: broader coverage, deeper insights, and faster updates. Providers can realistically expand into topics or markets that were previously too resource-intensive to cover. As a bonus, these efforts may also lead to cost savings, as AI can assist and in some cases replace previously manual tasks.

Double-down on proprietary data and insights.
Information services provide a mix of assets to their customers: proprietary data, content that is more general in nature, and software and workflow tools. As AI capabilities advance, the moat for proprietary data remains strong. While AI companies are able to replicate public datasets and automate general workflows, they cannot do the same for the proprietary content and benchmarks that have been refined, written into contracts, and embedded into workflows over decades. Information services providers, therefore, should continue to invest in refining and protecting the unique data assets that have always been one of the industry’s core strengths.

The AI era will no doubt bring changes to the information services industry that we cannot predict today. But providers that take steps to solidify their sources of advantage will continue to be critical partners for professional services firms in the years to come.

We thank our colleagues Lauren Hawkins, Fergus Jarvis, Rocio Larraguibel, Jake Barone Malis, and Tim Nolan for their contributions to this article.