Inbenta Holdings

Inbenta Holdings

Jordi Torras launched Inbenta in Barcelona in 2005, commercializing natural-language processing research that predated the current generative-AI wave. The company later relocated its headquarters to Foster City, California, while maintaining European operations in Paris and Madrid. Revenue is enterprise-software licensing, not venture-scale asset management; Inbenta is a technology product company, not a fund, though its history of $40 million in disclosed equity funding from early-stage backers marks it as a capital-backed operator. Inbenta sells a suite of conversational AI modules — chatbots, knowledge management, and semantic search — deployed across customer service, IT support, and HR use cases. Unlike competitors that layer prompts on large language models, Inbenta's core differentiator is its symbolic AI engine, which uses lexical semantics and a proprietary lexicon of over 10 million word relationships to understand queries without training on customer data. The company reports implementations at Amadeus, The Cheesecake Factory, and Kyocera, among others. Geographically, Inbenta serves North America, Europe, and Latin America, with a São Paulo office supporting Brazilian enterprise demand. Inbenta employs a global team with primary hubs in the Bay Area, Paris, and Madrid. The firm's disclosed equity funding includes a $12 million Series A in 2016 and a $25 million Series B in 2018 led by Semillero Partners, a growth-equity firm. In May 2018, Inbenta acquired EU-Calderon, a Barcelona-based digital marketing agency, to expand its technology footprint into AI-driven personalization. The firm's go-to-market architecture combines direct enterprise sales with a growing channel-partner network of systems integrators. What separates Inbenta structurally from typical enterprise SaaS is its refusal to cede query understanding to a third-party large language model — the company owns its entire NLP stack, from the lexical database upward, making it one of the few independent AI firms still fielding a proprietary alternative to the GPT ecosystem for mission-critical enterprise deployments where data residency and deterministic answers matter.