Venture investors have spent the last two years sorting signal from noise in artificial intelligence. The early rush to back foundational models has expanded into something more nuanced, with attention shifting toward startups building the layers around those models: context infrastructure, autonomous agents, vertical applications, and physical AI systems that could make generative technology operational at scale. For many investors, the question is no longer simply who has the best model, but who can make AI dependable enough for enterprise and industrial use.<\/p>\n
That shift is also changing what makes a startup attractive. Rather than pursuing companies that promise generalized disruption, many VCs are showing growing interest in businesses addressing narrower but potentially foundational gaps: reducing hallucination risk in enterprise agents, turning compliance into software, enabling digital workers, or extending AI beyond software into robotics. The startups below reflect several of those themes and offer a glimpse into where early-stage conviction is beginning to form.<\/p>\n
Naboo<\/p>\n
Founded around a premise that retrieval alone is insufficient for enterprise-grade AI, Naboo<\/a> is developing what it calls an intelligent context layer for agents and developers. Its platform connects code repositories, tickets, pull requests, documentation, communication tools, and metadata into a continuously updated semantic layer designed to help agents operate with a more complete understanding of enterprise environments. The company\u2019s argument is that context in modern organizations lives beyond code, and AI systems need access to those relationships if they are to execute reliably.<\/p>\n That framing resonates with investors watching enterprises move from AI experimentation toward production use. Naboo\u2019s focus on intent calculation, execution boundaries, and on-premises deployment also aligns with demand for systems that can meet security and compliance expectations. As agentic architectures evolve, startups positioning themselves as infrastructure rather than application layers are increasingly drawing attention.<\/p>\n Manus<\/p>\n Manus<\/a> sits at the intersection of no-code development, automation, and agent-driven workflows. The company enables users to launch business applications, create mobile apps, generate visuals, build presentations, and automate workflows without traditional engineering resources. Its broader ambition appears to be turning natural language into a universal interface for software creation.<\/p>\n That broad product surface may be one reason it has become a company many investors are watching. Manus is not narrowly positioned as another productivity assistant; it is making a larger case around AI-generated software as a category. For VCs interested in whether agentic tools can move from copilots into systems of creation, Manus offers a test case.<\/p>\n Bolt<\/p>\n Developer tooling has emerged as one of AI\u2019s most active categories, and Bolt<\/a> is gaining attention by positioning itself around coding agents rather than isolated coding assistants. Its platform brings multiple frontier agents into a unified interface, with an emphasis on handling larger-scale projects, reducing errors through iterative testing, and improving context management as software complexity grows.<\/p>\n That pitch reflects a broader investor thesis that the next wave of developer AI will be defined less by code generation and more by orchestration. Bolt\u2019s value proposition of helping product builders move from concept to deployed software while reducing operational friction places it in a crowded market, but also one where VCs continue to see room for breakout players.<\/p>\n 11x<\/p>\n If much of the AI conversation has centered on copilots, 11x<\/a> represents the more ambitious \u201cdigital worker\u201d narrative. The company develops AI workers designed to operate continuously across business functions, particularly in revenue and pipeline-oriented workflows. Rather than positioning automation as task assistance, 11x frames it as workforce augmentation.<\/p>\n That distinction matters to investors. Many see digital workers as one of the more consequential commercial categories emerging from generative AI, particularly if companies can show measurable productivity gains. 11x\u2019s emphasis on enterprise-ready deployment and embedded operational roles has helped place it within a growing class of startups testing whether AI can become labor infrastructure rather than software overlay.<\/p>\n Skild AI<\/p>\n