The 2025 AI reality check: What 1,100+ companies actually deploy vs. what they claim
Key findings at a glance from our 2025 AI Benchmark Study:
- Only 6% of companies have fully deployed agentic AI solutions on their public-facing websites – despite widespread claims about AI transformation
- 83% of AI leaders report “major” or “extreme” concern about generative AI progress – an 8X increase since 2023
- Cost concerns have increased 18X since 2023 while security worries tripled
- Just 37.5% support multiple languages on their websites – the most overlooked essential capability
Two years have passed since generative AI came onto the scene and entered our collective consciousness. During this period, we have witnessed impressive advancements from AI leaders like OpenAI and Anthropic especially when it comes to agentic AI and AI agents. And we’ve heard too-many-to-count pronouncements of ways the shiny new technology will change our work and our lives.
… Yet we’ve also seen a gap between what organizations say they’re going to implement and what they have actually deployed in their digital experiences.
That’s why we developed an entirely different approach to market intelligence for our annual Generative AI Benchmark Report. Rather than relying solely on self-reported data, we built an autonomous AI agent named Guydbot to evaluate AI for digital experience implementations directly on company websites. This Market Intelligence Agent has now assessed more than 1,100 companies across 45+ industry segments, documenting precisely which AI capabilities function in production environments today.
Our motivation extended beyond market research. We recognized that organizations needed an honest assessment to cut through industry noise and identify genuine opportunities for improvement. Because too many companies operate under assumptions about competitor capabilities that simply do not reflect reality.
By providing this transparent view, we aim to help real people at real companies make informed strategic decisions rather than reactive ones based on incomplete information.
The methodology behind the measurement
Traditional market research relies on surveys, interviews, and questionnaires. Organizations describe their intentions, outline their strategies, and tally their claimed successes. These approaches gather valuable insights about planning and sentiment, but they don’t reveal a whole lot about actual implementation quality. (For example, we’ve captured AI spend plans year-over-year, as outlined below.)
Our Market Intelligence Agent, Guydbot, operates differently. This agentic AI system functions as a sophisticated customer, systematically evaluating 24 distinct capabilities by:
- Testing search relevance and personalization effectiveness
- Assessing conversational commerce implementations
- Verifying functionality through direct website engagement
- Examining capabilities across industries from pharmaceuticals to automotive manufacturing
You can read about all 24 capabilities in our full report.
Rather than accepting claims about AI deployment, Guydbot verifies functionality through direct engagement with a website. The AI agent clicks around and tests search relevance, examines personalization effectiveness, and assesses conversational commerce implementations across industries ranging from pharmaceuticals to automotive manufacturing.
This is the first time an agentic AI system has conducted industry-wide capability benchmarking at this scale. The results provide unprecedented visibility into the gap between aspiration and execution in AI for digital experience deployments.
The AI Capability Matrix framework
Our analysis reveals that organizations cluster into four distinct groups based on their implementation of essential e-commerce capabilities and advanced AI functionality. This AI Capability Matrix provides a clear framework for understanding competitive positioning in AI for digital experience.
It’s also important to note that all company assessments are point-in-time analyses, and may change over time.
Achievers constitute 35% of evaluated companies. These organizations have mastered foundational capabilities while successfully deploying sophisticated AI features. Amazon, Tesla, and Microsoft exemplify this balanced approach, delivering compelling customer experiences through comprehensive implementation.
Builders represent 14% of the sample. Companies in this cohort excel at essential capabilities including dynamic facets, personalization, and product availability information. However, they maintain measured approaches toward emerging technologies. Neiman Marcus and Ford are examples of this careful progression from strong foundations toward advanced functionality.
Climbers comprise 10% of organizations. These companies pursue cutting-edge AI capabilities but have some gaps in fundamental requirements. Macy’s and Winnebago illustrate this innovation-focused approach, implementing conversational commerce and guided selling without fully addressing some of the basic e-commerce essentials.
Spectators account for 41% of evaluated companies. These organizations are still developing capabilities across the board. Rather than indicating failure, this positioning can actually reflect strategic decisions about organizational readiness, talent development, or market timing.
The implementation reality: A much-needed gut check
Guydbot’s assessment exposes significant discrepancies between the common FOMO-driven panic about AI adoption that dominates media coverage (and board rooms) and actual deployment realities. This represents an industry gut check that many organizations desperately need.
Here’s the reality: While 71% of companies have adopted some aspects of generative AI capability, only 6% have fully deployed agentic AI solutions for customer experiences. This statistic alone should give pause to executives who assume their competitors have rushed ahead with public-facing AI agents.
More striking, 65% of organizations lack fundamental capabilities necessary to support sophisticated agentic AI features in their digital experiences. This finding highlights why so many implementation attempts struggle: companies pursue advanced capabilities without establishing the foundational infrastructure these systems require to function effectively.
The data also reveals specific implementation gaps that directly impact business performance today. For example, only 37.5% of companies support multiple languages despite serving global markets.
This pattern exposes a logical playbook for catching up. Rather than investing in sophisticated natural language processing but leaving basic Spanish (or French, etc.) language support on the sidelines… or developing conversational AI agents while failing to provide essential product information, why not use the momentum from AI to also invest back into the less-expensive, low-hanging-fruit capabilities?
The principle is straightforward: meet customers where they are today while preparing for where they will be tomorrow. Advanced AI capabilities will indeed transform customer experiences in the near future, but organizations cannot afford to overlook accessible improvements that deliver immediate value.
“One for you, one for them,” as our CEO Mike Sinoway says.
Our research reveals that most organizations face similar challenges, suggesting that thoughtful, methodical approaches to AI deployment may prove more valuable than rushed implementations designed to close perceived gaps that may not actually exist.
The anxiety paradox
Why most companies feel behind (but aren’t)
We also ran a concurrent survey of more than 1,600 AI practitioners — and it revealed a concerning trend. 83% report “major” or “extreme” concern regarding generative AI progress. This represents an 8X increase in AI anxiety since 2023, when only 10.2% expressed similar anxieties.
This escalation reflects genuine challenges across multiple areas:
- Cost concerns: 18X increase since 2023
- Data security worries: Tripled
- Accuracy and reliability issues: 8X growth
This is how the paradox emerges: executives assume competitors have achieved substantial AI advances while simultaneously struggling with their own deployments. In reality, most organizations run into similar implementation challenges. This creates artificial urgency that often leads to rushed initiatives and poor outcomes.
The dual-track imperative
Our 2025 Generative AI Benchmark Report reveals that the highest-performing organizations follow a consistent dual-track approach. Rather than choosing between foundational capabilities OR advanced agentic AI features, they simultaneously accelerate essential implementations while exploring autonomous frontiers in their digital experiences.
Essential capabilities deliver 2X the conversion impact of advanced AI features in isolation, challenging common assumptions about AI implementation priorities.
Industry-specific insights
Different sectors exhibit distinct implementation patterns that reflect their unique requirements and customer expectations. B2C businesses have embraced AI somewhat more aggressively, with 41% qualifying as Achievers compared to 31% of B2B organizations.
Leading industries by AI implementation
- The software industry leads in agentic AI deployment, with 31% implementing technical interrogation capabilities. Microsoft, IQVIA, and CDW show sophisticated conversational interfaces and decision support tools that enhance customer engagement while reducing support costs.
- Online apparel retailers excel at conversational commerce, with Revolve, ThredUp, and Zalando implementing agentic AI-powered size recommendations and shopping assistants. These implementations address specific industry challenges around fit, returns, and product discovery.
- Travel companies focus on customized promotions and content enrichment. Airbnb pursues comprehensive travel concierge functionality through agentic AI, while Marriott’s “RenAI” provides local guidance for Renaissance Hotels. These implementations transform customer service from reactive support to proactive assistance powered by AI for digital experience optimization.
The path forward: From insight to action
The evidence from our 2025 Generative AI Benchmark Report suggests a measured approach to AI implementation yields better results compared to aggressive pursuit of cutting-edge capabilities without foundational strength. Organizations should prioritize capabilities like hybrid search technology, personalization systems, and intelligent product discovery before pursuing agentic AI agents and natural language interfaces. (Or, at the same time.)
This recommendation does not diminish the importance of agentic AI capabilities for digital experience enhancement. Rather, it acknowledges that these advanced features deliver even more value when built upon sturdy foundations. For example, companies implementing conversational commerce without effective search relevance can hurt the experience. Organizations deploying agentic AI chatbots without proper product data integration generate inaccurate responses that erode trust.
Most importantly, our findings provide actionable intelligence for strategic planning. Rather than operating under assumptions about market conditions, organizations can now benchmark their capabilities against verified industry data by downloading our 2025 Generative AI Benchmark Report. This transparency allows for more informed resource allocation, realistic timeline planning, and strategic positioning based on actual competitive landscapes rather than perceived ones.
Beyond the measurement: Empowering strategic decisions
This study represents more than market research. It establishes a new standard for evaluating AI implementation that extends beyond claims and intentions to actual customer impact through AI for digital experience measurement. Guydbot’s agentic AI assessment capability provides ongoing visibility into market changes, competitive positioning, and implementation effectiveness.
The ultimate goal extends beyond documenting current conditions. We created this comprehensive analysis to empower organizations with the strategic intelligence needed to make informed decisions about AI investments.
As agentic AI systems become more sophisticated, this approach will become increasingly valuable for organizations seeking objective evaluation of their digital experiences. Our research provides the foundation for these informed decisions, helping real people at real companies confidently explore complex technology landscapes.
Download the complete 2025 State of Generative AI in Global Business report to explore detailed industry analyses, capability assessments, and strategic recommendations for your organization’s AI journey.
