By admin 
The persistent "last mile" challenge of digital transformation, a terrain historically strewn with forgotten PDF manuals and overlooked training documents, is now facing a seismic shift with the advent of agentic AI. For years, organizations have invested heavily in sophisticated enterprise software such as SAP and Salesforce, only to witness employees grappling with basic system navigation. This pervasive issue now presents a dual imperative as the era of agentic AI dawns: companies must equip their human workforce with the skills to effectively collaborate with AI, while simultaneously imbuing AI agents with the capacity to navigate the intricate and often labyrinthine interfaces of modern enterprise systems.
Emerging as a promising solution, a novel approach is gaining traction among forward-thinking businesses: leveraging screen recordings and detailed tutorials of employees performing critical enterprise tasks, from generating a new support ticket to processing an invoice. This visual data is then used to train AI agents to meticulously replicate these workflows. This week, the startup Standard Intelligence garnered significant attention on X (formerly Twitter) with an early demonstration of an open-ended application of this methodology, encompassing both physical and digital realms. However, the reality is that established players are already tackling this enterprise knowledge infrastructure crisis head-on.
Case in point is Guidde, an Israeli startup that emerged during the video-centric landscape of the COVID-19 pandemic. Today, Guidde announced an oversubscribed $50 million Series B funding round, led by PSG Equity. This substantial investment is earmarked to address the critical knowledge infrastructure deficit that hinders both human and AI adoption of enterprise technologies. Instead of relying on static PDF manuals, Guidde champions the concept of "Video Ground Truth"—a rich, dynamic stream of data captured from experienced human users as they expertly navigate complex software applications. This funding round signifies a pivotal re-evaluation within the tech industry, reframing documentation not as a static byproduct of work, but as the essential telemetry required to train the next generation of autonomous digital agents.
Technology: From Granular Video Capture to Sophisticated World Models
At its core, Guidde operates as an AI Digital Adoption Platform (ADAP). Its technological prowess, however, lies in the sophisticated processes that occur behind the scenes during the recording phase. Guidde transcends mere pixel recording; it meticulously captures every click, scroll, and latent interaction with the underlying HTML page. This includes subtle pauses, precise scroll depths, and crucial corrections made by a human user in response to system lag or unexpected behavior. This granular telemetry transforms raw video footage into a high-fidelity Vision-Language-Action (VLA) training dataset.
Furthermore, Guidde’s proprietary Magic Redaction technology automatically obscures sensitive information, such as passwords and credit card numbers, during the capture process. This ensures that all recorded materials remain secure and compliant with stringent regulations like HIPAA. "Every time you click a button, you drag-and-drop, you scroll, you type, we gather the interaction… all of it, we do cleanse it—there’s no private information," explained Guidde co-founder and CEO Yoav Einav in an exclusive interview. This meticulous approach ensures data integrity and privacy.
The platform’s sophisticated engine captures the underlying metadata and Document Object Model (DOM) changes, perfectly synchronized with each video frame. The true differentiator, however, lies in the depth of the telemetry captured beneath the surface of the visual recording. This rich metadata collectively constructs a "digital world model" of enterprise software environments. Given that each enterprise utilizes a unique amalgamation of applications and internal processes, Guidde is actively building a formidable data moat. This moat empowers enterprise AI agents to reason through even legacy user interfaces with the same spatial awareness and contextual understanding as a human user, thereby ensuring that automation solutions are not confined to lab demonstrations but are genuinely functional in production environments. For human employees, these recordings serve as intuitive tutorials; for AI agents, they function as a high-fidelity blueprint of the interface. This enables agents to "see" and intelligently interpret complex user interfaces in a manner akin to human perception, effectively solving the persistent "last mile" of automation where agents have historically faltered due to a lack of specific enterprise context and in-situ usage data. In essence, Guidde is pioneering a "self-driving car" for computer interactions, analogous to Waymo’s impact on autonomous driving.
Product: Three Pillars of Guidd-ance
Guidde’s platform has strategically evolved into three distinct product offerings, designed to scale seamlessly with an organization’s digital adoption maturity:
-
Magic Capture: This core feature enables users to effortlessly record any digital workflow. It automatically detects user actions, contextualizes them with underlying system data, and generates structured narrative scripts, actionable insights, and professional-grade AI voiceovers. This dramatically reduces the time and effort required to create instructional content.
-
Video Ground Truth: This represents the rich, multimodal data asset generated by Magic Capture. It serves as the authoritative source of truth for training AI agents and for providing contextual, in-flow guidance to human users. This data includes not only visual cues but also the precise sequence of actions, timing, and system responses.
-
AI Training and Digital Adoption: This pillar encompasses Guidde’s ability to leverage the Video Ground Truth data to both train AI agents and deliver in-the-moment guidance to human users. This dual functionality is key to unlocking widespread AI adoption and enhancing human productivity.
Training Humans on AI and AI on Humans
Perhaps the most intriguing and forward-thinking aspect of Guidde’s strategic growth is its dual-purpose mission. "We’re the only platform that trains both humans and agents," Einav emphasized. As organizations increasingly deploy advanced AI tools like Microsoft 365 Copilot or ServiceNow agents, they frequently encounter a significant proficiency gap. One of Guidde’s largest enterprise clients, for instance, revealed that despite investing over $1 million annually in a sophisticated AI tool, "nobody knows how to use them because they did like a 30-minute training session, and then that’s it." Guidde effectively bridges this gap by delivering "bite-sized" video tutorials directly within the user’s workflow, ensuring learning is contextual and immediately applicable.
Simultaneously, these meticulously crafted videos serve as the foundational training material for AI agents themselves. Large language models and foundation models, such as Gemini or GPT-4, can often exhibit "hallucinations" or perform erratically when tasked with highly specific enterprise workflows. This is largely because they were not trained on the unique, internal "vanilla workflows" that characterize private enterprise systems. Guidde provides the crucial "starting point," the essential "metadata," and the precise "x, y coordinates of the button" that an AI agent requires to successfully complete an action without becoming disoriented or failing.
The Multimodal Advantage
To maintain its exceptional level of accuracy and reliability, Guidde employs a sophisticated multimodal infrastructure. The system does not rely on a single AI model; instead, it utilizes a dynamic "fleet" of models that continuously evaluate and cross-reference each other’s outputs. This collaborative approach enables Guidde to consolidate a traditionally fragmented stack of six or seven disconnected tools—such as Loom for screen capture, Adobe Premiere for video editing, 11Labs for text-to-speech narration, and Synthesia for AI-generated avatars—into a single, integrated, AI-native platform. "We basically pack everything for you," Einav stated, "and automate the entire process based on your brand guidelines." This end-to-end automation streamlines content creation and ensures brand consistency.
Video-First Origin Story
The genesis of Guidde is rooted in a frustration familiar to many product leaders and entrepreneurs. Prior to co-founding Guidde, Einav and co-founder Dan Sahar dedicated years to mastering video traffic analysis at Qwilt, a company they established in 2010 to study user viewing habits on platforms like Netflix and Disney+. When the COVID-19 pandemic accelerated remote work trends, they recognized a significant opportunity to apply their extensive video expertise to the challenges of the modern workplace. They observed that concise video explainers could dramatically increase free-to-paid account conversions by as much as 30%. However, the manual effort involved in creating such videos proved to be an unsustainable bottleneck.
In a candid interview, Einav recalled the "tedious work" of the pre-Guidde era: "My team in Israel were creating the content, someone in the US with a US accent was doing the narration, someone in the marketing team would write the script… and someone in the enablement team would do the edit." This highly fragmented workflow meant that producing a single instructional video could take two to three weeks. "And then two weeks later, the product changes, and you need to redo it from scratch," Einav lamented. Guidde was conceived and built specifically to collapse this protracted cycle into mere seconds. By automating the "Magic Capture" of any workflow, the platform instantly generates a structured narrative script and a professional AI voiceover. This eliminates the editing bottleneck, effectively transforming subject matter experts into highly efficient "training powerhouses."
Licensing and Market Impact
Guidde’s pricing structure reflects its evolution from a niche utility to an indispensable component of enterprise infrastructure. The platform offers tiered licensing options designed to accommodate businesses of varying sizes and needs, ranging from individual team licenses to enterprise-wide deployments. This flexible model ensures accessibility and scalability.
The tangible impact of Guidde’s platform is already demonstrably evident in key performance metrics. Customers are reporting an impressive 41% reduction in video creation time and a significant 34% decrease in inbound support tickets. For large enterprises like Emerson, this translates into dramatically accelerated guide creation, with reported speed-ups of 40-60%. Support teams, in particular, are realizing substantial efficiencies, with the capability to offload an estimated 80% of their ticket volume to AI agents—provided, critically, that those agents are equipped with the necessary, up-to-date content. "The agent without the content is useless," Einav cautioned, highlighting the prevalent issue of enterprise documentation being either severely outdated or entirely absent.
Community and Industry Early Reception
Guidde has already garnered a substantial user base, boasting approximately 4,500 enterprise customers. The company aims to further expand this number significantly with the capital secured from its latest funding round. Leaders in support and operations have been particularly vocal in their praise for the platform’s user-friendliness and effectiveness. Christopher Cummings, VP of Client Experience at DocNetwork, highlighted Guidde’s ability to provide "quick, personalized video responses to customer questions," a capability that significantly enhances customer satisfaction. Similarly, Wren Cotrone, a Director of Customer Support, noted, "Once you set the branding the way you want, you can really zoom through this stuff," underscoring the platform’s efficiency once configured.
Ronen Nir, Managing Director at PSG, eloquently summarized the investment thesis, stating, "Guidde is solving one of the biggest blockers to successful AI adoption: the knowledge infrastructure." This sentiment underscores the critical role Guidde plays in enabling organizations to fully leverage the transformative potential of AI.
Why This Matters Now
The ongoing paradigm shift from text-only Large Language Models (LLMs) to sophisticated agentic video intelligence represents a defining trend for the immediate future. Guidde’s Series B funding round serves as a clear indicator that the "ground truth" for enterprise AI agents will increasingly originate from raw video observation rather than static, often outdated, textual documentation. By meticulously capturing how work is performed across tens of millions of unique workflows, Guidde is accumulating a dataset of unparalleled breadth and depth—a proprietary asset that few other entities can rival.
As Yoav Einav articulated, "It starts with humans in the loop, and over time moves toward full autonomy." For the contemporary enterprise, the operational map is no longer a static document; it is rapidly evolving into a dynamic, living, breathing video intelligence layer. This layer serves as the essential guide for both the human workforce and the increasingly sophisticated AI agents that support them, ushering in a new era of intelligent automation and digital adoption.