HubSpot’s New AI Engine Optimization Tool Falls Short, Highlighting a Broader Industry Trend of Underperforming AI Features

HubSpot, a long-standing pillar of the SaaS and B2B technology landscape, is facing scrutiny over the efficacy of its newly launched AI Engine Optimization (AEO) tool. While the company, a favorite within the SaaStr AI community, has been lauded for its foundational contributions to the industry, its recent foray into AI-powered optimization has revealed a significant gap between its offerings and the rapidly evolving capabilities of specialized, AI-native competitors. This development underscores a concerning industry-wide pattern: established B2B companies are releasing AI products and features that, while technically functional, often deliver only a fraction of the performance seen in leading point solutions. In the accelerated AI era of 2026, this "60% solution" approach is proving insufficient to capture market attention and drive growth.

The launch of HubSpot’s AEO tool, announced as part of a broader AI initiative, was met with anticipation. However, initial user experiences, including that of Jason Lemkin, founder of SaaStr, suggest that the tool, while technically operational, falls considerably short of its more agile, agentic competitors. The core issue, as articulated by industry observers, is the expectation that in 2026, AI solutions must offer near-best-in-class performance to justify their adoption and monetization. A product that is merely "60% as good" as dedicated alternatives, especially when coupled with a recurring subscription fee, struggles to find traction in a market increasingly defined by rapid innovation and fierce competition.

A Closer Look at the HubSpot AEO Tool Experience

The user experience with HubSpot’s AEO tool, as detailed by early testers, paints a picture of functionality without deep utility. Upon utilizing the tool, users were presented with a dashboard displaying data points that lacked clear actionable insights. Critically, the tool failed to provide specific recommendations for improvement or cite sources that offered novel information, often reiterating already known data. This lack of practical guidance was compounded by an immediate prompt to upgrade to a paid tier, priced at $50 per month for additional prompts. This pricing, observers note, is steeper than that of many specialized agentic products that offer more robust functionality.

The central question arising from this experience is the perceived value proposition. While a free, included feature might be acceptable for its basic utility, the tool’s performance as a standalone product, particularly given its cost, is being called into question. The competitive landscape in AI is characterized by aggressive development and swift iteration. Companies like Replit, Lovable, Reve, Higgsfield, Opus, and Gamma have established themselves as leaders in their respective niches by delivering AI-native solutions that excel in specific domains. These specialized tools have benefited from months of compounding improvements, ingesting vast datasets, and refining their algorithms, widening the gap between "good enough" and "best-in-class."

A particularly striking aspect of the initial HubSpot AEO tool rollout was its assessment of the SaaStr.ai website. The tool reportedly assigned a score of 0% to the site and provided no recommendations for improvement. While HubSpot representatives suggested a potential misunderstanding of the tool’s capabilities, the absence of concrete, actionable advice in an AI-grade product in 2026 is seen as a critical failing. This lack of guidance renders the tool less effective for users seeking tangible improvements to their online presence or AI engine optimization.

The "60% Problem": An Industry-Wide Challenge

The limitations observed in HubSpot’s AEO tool are not an isolated incident but rather indicative of a broader trend across the B2B software sector. Numerous vendors are rushing to integrate AI features to meet market demand and avoid appearing technologically stagnant. However, many of these efforts result in "checkbox AI" – features that fulfill the requirement of having AI but do not offer a truly superior user experience or problem-solving capability compared to existing, dedicated solutions.

This strategy may have been viable six to nine months prior, when the AI landscape was less mature and buyers were more forgiving of nascent AI functionalities. The rapid advancements in Large Language Models (LLMs) throughout 2025 significantly raised the bar for what constitutes effective AI. The window for releasing moderately functional AI features has now closed. Dedicated AI-native point solutions have had ample time to refine their offerings, often achieving significant revenue growth and market dominance. For instance, Replit and Lovable in the coding space, Reve for image generation, Higgsfield and Opus for short-form video, and Gamma for presentations, have set high benchmarks. These companies have benefited from continued data ingestion, iterative development cycles, and deeper integration into user workflows, solidifying their positions as category leaders.

The widening chasm between "60% good enough" and "best-in-class" means that established players introducing AI features that merely approximate existing functionalities risk being relegated to supplementary roles, if they are adopted at all. Customers, particularly those willing to pay for advanced solutions, are unlikely to invest in tools that do not demonstrably outperform specialized alternatives.

The Rise of Rapid Prototyping: A Competitive Response

In response to the perceived shortcomings of existing AI tools, some industry figures have resorted to building their own solutions. Jason Lemkin, in an experiment to illustrate the point, developed his own AI Engine Optimization (AEO) analyzer using Replit. This initiative, completed within a matter of hours, aimed to create a tool that provided actionable recommendations, a feature conspicuously absent in the HubSpot offering.

The self-built AEO tool, accessible via a public link, reportedly scored the SaaStr.ai website at C+ (67/100) and provided specific, actionable prompts for improvement. For example, it identified deficiencies in structured data implementation (JSON-LD, Microdata), heading hierarchy, table content, and content freshness signals. The tool then generated ready-to-use prompts that could be integrated into AI coding assistants like Replit, Lovable, or Cursor, or directly implemented into content management systems like WordPress and Shopify. This level of detail and direct applicability stands in stark contrast to the 0% score and lack of recommendations reportedly provided by HubSpot’s tool.

This rapid development cycle, where a non-engineer can build a functional alternative to a commercial AI product in under a day, poses a significant threat to companies relying on incremental AI feature releases. The ease with which such competitive tools can be created and deployed suggests that "60% solutions" have a rapidly diminishing shelf life, potentially measured in weeks rather than years. This phenomenon is not unique to AEO tools but is occurring across various B2B categories, inviting a wave of more capable, often free, alternatives.

While Lemkin acknowledges the value of purchasing rather than building, his experiment highlights the market’s demand for utility and actionable insights. The existence of such rapidly developed, competitive tools suggests a shift where innovation speed and genuine problem-solving are paramount, potentially leading users to opt for readily available, effective solutions over expensive, underperforming ones.

Figma Make: A Case Study in Missed Opportunity

The dynamic of established players entering crowded AI markets late and with less effective products is vividly illustrated by the case of Figma Make. Launched into the burgeoning "vibe coding" or AI-assisted website design space, Figma Make’s performance has been met with disappointment, particularly from a company with Figma’s design pedigree and distribution power.

During a test in March 2026, the Figma Make tool was tasked with redesigning the SaaStr AI website. The output was described as generic, featuring commonplace gradients and placeholder content, reminiscent of earlier AI startup aesthetics from 2025. More critically, the tool failed to extract content from the existing website and, in some instances, hallucinated the nature of the site itself, resulting in a design that was demonstrably worse than the current iteration. This performance is particularly noteworthy given that leading AI coding platforms can reportedly handle such tasks with ease in 2026.

Figma Make’s release timing is seen as a significant factor in its struggles. Had it been launched in the summer of 2025, when the vibe coding category was nascent and companies like Lovable were still emerging and Replit was solidifying its AI capabilities, Figma’s brand recognition and extensive user base could have provided a substantial advantage. At that time, the market was less saturated, and a design-tool-turned-app-builder could have captured significant territory.

However, the market has evolved dramatically in the intervening months. By early 2026, dedicated AI coding platforms had experienced explosive growth. Lovable, for example, had reached $400 million in Annual Recurring Revenue (ARR) by March 2026, adding $100 million in a single month. Replit was also reporting ARR exceeding $400 million, with ambitions to reach $1 billion by year-end, and had secured significant funding at a substantial valuation. Combined, these specialized platforms represent a vibrant and rapidly growing market, estimated to be well over $1 billion in total ARR, generated within approximately 12-18 months.

Figma’s attempt to monetize its AI features through credit add-ons, initiated in March 2026, has so far yielded nominal revenue, described by management as "measured" and a "rounding error" compared to the financial performance of dedicated AI coding tools. This situation exemplifies how a product released into a mature and highly competitive market, even by a well-established player, can struggle if it doesn’t offer a best-in-class experience. Figma Make, by failing to innovate beyond generic mockups and lacking fundamental content integration capabilities, missed its opportune window, becoming a "pack-in" feature rather than a market leader.

Underlying Reasons for the "60% Solution" Phenomenon

Several factors contribute to the prevalence of underperforming AI features being shipped by established B2B vendors:

• Legacy Systems and Infrastructure: Older technological stacks and complex legacy systems can impede the rapid integration and optimization of cutting-edge AI models. Adapting these systems to harness the full potential of AI requires significant engineering effort and time.
• Risk Aversion and Incrementalism: Large, established companies often prioritize stability and predictability over radical innovation. This can lead to a preference for incremental feature additions that integrate with existing products, rather than developing entirely new, AI-native platforms that might carry higher development risks but offer greater rewards.
• Organizational Inertia and Culture: Established corporate cultures can sometimes be slow to adapt to the fast-paced, experimental nature of AI development. Decision-making processes, team structures, and incentive models may not be aligned with the rapid iteration cycles required to compete effectively in the AI space.
• Misunderstanding of AI’s Potential: Some organizations may still be in the early stages of understanding the transformative capabilities of AI, viewing it as an add-on rather than a foundational technology. This can lead to a focus on superficial AI features that don’t leverage the full power of the technology to solve core user problems.

The Permanently Elevated Bar in AI Development

The current AI landscape is fundamentally different from previous technological shifts, such as the early days of mobile or cloud computing. The speed at which AI capabilities are improving is unprecedented. In the mobile era, a mediocre app released in 2009 had several years to iterate before market consolidation. In contrast, AI advancements are so rapid that a feature considered functional in April could be rendered obsolete by October of the same year.

This accelerated improvement curve means that products that were competitive six to nine months ago are now merely supplementary "pack-ins." The "60% solutions" shipping today face an even more precarious future, potentially becoming irrelevant by the end of the year.

For B2B vendors, the message is clear: if an AI feature cannot genuinely achieve best-in-class performance, or at least come remarkably close within a specific use case, it is often more detrimental to ship it than not to ship it at all. A subpar AI feature can damage a company’s brand, leading its most valuable customers to conclude that the platform "doesn’t really do AI." These customers, who are often the most willing to pay for advanced solutions, will then seek out and purchase dedicated point solutions that offer superior functionality.

The market has moved beyond the phase of evaluating AI products on a curve. The expectation is now for excellence. Companies must either deliver truly groundbreaking AI capabilities or refrain from introducing features that fail to meet the high standards set by the market. The era of accepting "good enough" in AI has definitively ended, replaced by a demand for innovation that drives genuine value and solves problems with unparalleled efficiency.