Clipto Reaches $250M Valuation With Local-First AI Search for Video Libraries

The Abundance Problem

Henry Kang spent 2006 teaching robots to remember where they had seen objects. Twenty years later, the fundamental challenge hasn't changed: how do you find something when you have too much to look through? The difference now is that the deluge isn't happening in a lab; it's sitting on every creator's external drive, every lawyer's desktop, every researcher's laptop.

Clipto, the startup Kang founded in 2023, has just closed a $15 million equity round at a $250 million post-money valuation. Investors include HSG (formerly Sequoia China), GL Ventures, EnvisionX Capital, Palm Drive Capital, Hans Tung, Lu Zhang, and 522 Ventures. The company operates from San Francisco with engineering and product teams in Singapore and Hong Kong, a footprint that reflects both the capital flows and the talent density across the Pacific tech corridor.

At its core, Clipto indexes video, audio, images, meetings, and documents stored locally on a user's machine. Instead of manually hunting through nested folders or scrubbing through hours of footage, users describe what they're looking for in natural language or allow AI assistants like ChatGPT and Claude to retrieve the material on their behalf. The entire process runs on-device; no files leave the user's computer unless explicitly authorized.

Kang told us the company reached $15 million in annual recurring revenue at the start of 2026 and remains net-income profitable. More than 30 million people have tried Clipto since launch, and the company now counts hundreds of thousands of paying subscribers, though it declined to share more granular retention or average revenue figures. Kang did note that a substantial share of customers remain subscribed beyond two years, a metric that matters in a market where many AI tools struggle to prove long-term utility.

From Wardrobe AI to Zenvideo to Search

Kang's path to Clipto traces through two earlier startups. The first applied computer vision to closet management, suggesting outfits based on what users owned. The second, Zenvideo, focused on simplifying video production and was acquired by Tencent in 2020. Between those ventures and his PhD work at Carnegie Mellon, Kang has spent nearly two decades wrestling with the same core question: how do you make large, unstructured datasets navigable?

The insight that led to Clipto emerged from that Tencent exit. While much of the AI industry was racing to build better generative models, Kang and several members of his previous founding team saw a different bottleneck. Content creation was getting easier, but content management was not keeping pace. Video creators in particular were drowning in footage scattered across multiple drives, with no efficient way to locate specific moments without manual review.

Clipto initially targeted that creator segment. Today, however, video producers represent only about a quarter to a third of the user base. The rest spans legal professionals reviewing depositions, doctors referencing recorded consultations, marketers repurposing campaign assets, HR teams searching interview recordings, and academics managing research footage. The product turned out to solve a broader problem than Kang had initially scoped.

Local Indexing in an MCP World

Two weeks before the funding announcement, Clipto added support for the Model Context Protocol, a standard that allows AI applications to query external data sources in a structured way. For Clipto, MCP support means that when a user grants permission, an AI agent can search indexed files within defined parameters without requiring manual export or upload.

Kang emphasized that all indexing and retrieval happens locally. Files are not sent to the cloud for processing, and access requires explicit user authorization each time an AI tool queries the index. This architecture addresses two concerns at once: latency and privacy. Local processing keeps search fast even when working with terabytes of video, and it sidesteps the data-residency and compliance issues that come with cloud-based indexing, particularly for legal, medical, and enterprise users.

The company's infrastructure challenge is less about building massive data centers and more about optimizing models to run efficiently on consumer-grade hardware. That requires careful trade-offs between model size, inference speed, and accuracy, especially when indexing high-resolution video in real time.

Competing With the Platforms

Clipto is entering a market where the largest players already control much of the content ecosystem. Adobe's Premiere offers AI-powered search within its video editing suite. Apple Photos and Google Photos let users find images and clips using natural-language queries. Microsoft is embedding similar features across its productivity stack.

Kang's counterargument is scope. Adobe's tools work within Adobe's universe. Apple and Google index files stored in their respective clouds. Clipto, by contrast, searches across video, audio, images, and documents regardless of where they live on a user's machine, and it integrates with multiple AI assistants rather than locking users into a single platform.

The question is whether that cross-platform, local-first approach is compelling enough to sustain a standalone product, or whether search will inevitably get bundled into the operating systems and creative suites people already use. At DailyTechWire, we've tracked similar debates in note-taking, password management, and backup software. In each case, a handful of independent tools survived by serving power users and enterprise customers who needed features the platform defaults didn't provide.

Clipto's bet is that the same dynamic will play out in media search. The company is profitable, which buys it time to find out. But the $250 million valuation suggests investors believe the market is large enough to support a dedicated product, at least for the segment of users whose content libraries have outgrown the tools built into their workflow software.

What the Funding Buys

The new capital will go toward two areas: model optimization and agent integrations. On the model side, Clipto needs to continue refining the AI that powers local indexing so it can handle larger libraries and more file types without requiring users to upgrade their hardware. On the integration side, the company is expanding the number of AI agents and productivity tools that can query Clipto's index through MCP and similar protocols.

Kang's long-term vision is less about replacing existing workflows and more about becoming the connective layer between a user's content and the AI tools they already use. If that works, Clipto becomes infrastructure: invisible, essential, and difficult to displace. If it doesn't, the features it pioneered will likely end up as checkbox items in the next OS update.

With just over 20 employees split between the Bay Area, Hong Kong, and Singapore, the company remains lean. That's both a risk and an advantage. It limits how quickly Clipto can expand, but it also means the company can remain profitable while it figures out whether the market it's building for is as large as its valuation implies.

The abundance problem Kang identified is real. Whether it's large enough to support a $250 million company that does one thing well, rather than a feature inside a product that does many things adequately, is the question the next two years will answer.