Clipto has raised $15 million at a $250 million post-money valuation for software that lets people search terabytes of their own video, audio, images and documents simply by describing what they are looking for. TechCrunch broke the news on 31 August 2026, and the figures behind the three-year-old San Francisco startup are unusual for an AI company of its age: roughly $15 million in annual recurring revenue, profitability on a net-income basis, more than 30 million users since launch and a team of just over 20 people.
The round backs a quieter thesis than most of the generative AI boom. Founder and CEO Henry Kang argues that the defining problem of the AI era is not a shortage of content but a glut of it — hours of footage sitting unused on hard drives because nobody can find the right moment. Clipto’s answer is an on-device index built on computer vision and speech recognition, plus a Model Context Protocol connector that lets ChatGPT, Claude and other assistants query that index on the user’s behalf.
This article breaks down what Clipto actually does, the funding and revenue numbers, who invested, how the pricing works, where the product sits against Adobe, Apple, Google and TwelveLabs, and what the “memory layer” idea means for businesses drowning in their own media. The large language model boom created the demand for tools like this one; Clipto is betting that finding things is now worth as much as generating them.
Table of contents
- What Clipto Actually Does
- The Clipto Funding Round in Numbers
- Who Invested in Clipto
- From Robots at Carnegie Mellon to Clipto: Henry Kang’s Path
- Clipto Pricing: What the Pro Plan Costs
- How Clipto Compares With Adobe, Apple, Google and TwelveLabs
- What Clipto’s “Memory Layer” Means for Businesses
- Frequently Asked Questions
- References
What Clipto Actually Does
At its simplest, Clipto indexes the videos, audio files, images, meeting recordings and documents already sitting on a user’s computer. Instead of scrolling through folders or scrubbing a timeline, the user types a plain-language description — “the clip where the chef plates the dessert” — and jumps straight to the matching moment. The same index can be handed to an AI assistant, so a user can ask ChatGPT or Claude to find the file rather than searching for it themselves.
Semantic search across terabytes of video
The company’s own description of the launch bundles several capabilities into one product: video understanding, semantic search, transcription, summaries and structured video notes, all working across “terabytes of video and audio using natural language”. Results trace back to the exact timestamp in the source file, which is the feature working editors care about most. The Pro tier adds automatic people tagging, support for more than 99 languages and analysis of individual videos up to six hours long.
Alex Grubb, a UK restaurant photographer quoted in the announcement, describes the workflow change bluntly: “The hardest part was never shooting, it was digging through hard drives to find one specific moment. Clipto tags everything automatically, so I just type a keyword and jump straight to the shot, down to the exact timecode.”
On-device processing and privacy
The architectural choice that separates Clipto from most AI search tools is where the work happens. According to TechCrunch, “all of this processing runs locally on the user’s device without requiring cloud services”, and the company markets the desktop product as “privacy first”, with users able to keep every search on-device. That is why the system requirements read more like a video-editing suite than a web app: a Mac with Apple Silicon M1 or later and at least 16GB of memory, or a Windows PC with 12GB or more of RAM. The iPhone and Android apps, by contrast, are cloud-powered.
Integrations: Premiere, DaVinci Resolve and MCP
Clipto ships plugins for Adobe Premiere Pro and DaVinci Resolve, so a search can happen without leaving the editing timeline. The newest integration is the one investors appear to be paying for. Roughly two weeks before the funding announcement, Clipto added support for the Model Context Protocol, the open standard that lets AI applications call external tools. Access “requires the user’s active request and authorization”, and connected apps “can only retrieve information within the scope the user specifies”. Native Windows and Android applications were announced alongside the round.
The Clipto Funding Round in Numbers
The headline figures are straightforward, but the supporting numbers are what make the round notable. TechCrunch reported the round as all-equity, and neither the company nor its investors named a lead.
| Metric | Figure | Source |
|---|---|---|
| Amount raised | $15 million, all-equity | TechCrunch, company announcement |
| Post-money valuation | $250 million | TechCrunch, company announcement |
| Annual recurring revenue | About $15 million at the start of 2026 | TechCrunch |
| Profitability | Profitable on a net-income basis | TechCrunch |
| Users since launch | More than 30 million | TechCrunch, company announcement |
| Paying subscribers | “Hundreds of thousands”; exact count not disclosed | TechCrunch |
| Headcount | Just over 20, across the Bay Area, Hong Kong and Singapore | TechCrunch |
| Lead investor | Not named | All sources |
Revenue, profit and headcount
A $250 million valuation on roughly $15 million of ARR works out at about 16.7 times revenue. That is a demanding multiple for a consumer subscription business, but well below the 30 to 60 times ARR that Kang told Mergermarket in April 2026 “high-growth AI software companies” can command. The chart below puts the three headline figures side by side.
Two details stand out. First, Clipto has raised only as much as it already books in a year, which is the opposite of the usual AI funding pattern. Second, the company is profitable with a team of about 20, which means the round is expansion capital rather than runway. Kang declined to give TechCrunch a precise subscriber count or average revenue per user, but said a significant share of customers stay subscribed for more than two years — a retention figure most consumer apps would envy.
What the money is for
The stated uses are specific: the AI models and computing infrastructure needed to run Clipto on ordinary consumer hardware, and integrations with more AI agents. Running video understanding locally on a laptop is an engineering problem as much as a product one, and the MCP work suggests the company sees AI assistants, not humans, as the next big consumer of its index.
Who Invested in Clipto
The investor list is a mix of Asia-focused growth funds and individual investors with long track records in consumer software: HSG (formerly Sequoia China), GL Ventures, EnvisionX Capital, Palm Drive Capital, 522 Ventures, and the individual investors Hans Tung and Lu Zhang.
Holly Zheng, founding partner at EnvisionX Capital, framed the bet in the announcement: “We believe AI’s next frontier isn’t just generating more content, it’s making sense of the enormous amount of content people already have. Clipto is tackling one of the hardest problems in AI: turning years of unstructured video into AI-readable memory while keeping data private and on-device.”
Earlier money and the April 2026 signal
The round did not come out of nowhere. In April 2026, Mergermarket reported that Clipto had raised roughly $20 million during 2025 from the same group of investors at a valuation above $250 million, and that Kang was planning a $50 million to $100 million raise within three to six months. The $15 million that actually closed is smaller than that ambition, and published totals for earlier rounds conflict — Tracxn lists $15 million in total across four rounds — so this article does not state a cumulative figure.
What the April report did establish is the shape of the business: Mergermarket put first-year ARR at around $10 million in 2023, said the company became profitable within three months, and described it as debt-free with Kang holding a majority stake.
From Robots at Carnegie Mellon to Clipto: Henry Kang's Path
Kang’s route to Clipto explains the product. The idea of a machine that remembers where things are has followed him for two decades.
| Year | Milestone | Detail |
|---|---|---|
| 2006 | Carnegie Mellon PhD research | Robots that recorded their surroundings, identified objects and remembered where to find them |
| Late 2000s–2010s | First startup, Peekabuy | AI that tracked the clothes in a user’s wardrobe and suggested outfits; sold to a strategic buyer |
| 2020 | Zenvideo acquired | Kang’s video-creation company bought by Tencent, according to Kang |
| 2023 | Clipto founded | Started in San Francisco with members of the Zenvideo founding team; roughly $10M ARR in year one |
| 2025 | Seed and pre-A funding | About $20M reported by Mergermarket from HSG, GL Ventures, EnvisionX, Palm Drive and 522 Ventures |
| August 2026 | MCP support launched | ChatGPT, Claude and other agents can query a user’s local media index |
| 31 August 2026 | $15M raised at $250M | Windows and Android apps announced in the same release |
Kang founded Clipto in 2023 with several members of his previous company’s founding team, and the business now runs from San Francisco with teams in Singapore and Hong Kong. His own summary of the opportunity, as told to TechCrunch: “The real insight is that in this AI era, we don’t have a content shortage. The opposite is true. We have too much content. We have too much video footage sitting on our computers that isn’t being used.”
Why creators are now a minority of Clipto users
Clipto was built for video creators, yet TechCrunch reports that creators now account for only about a quarter to a third of its users. The rest are lawyers, doctors, researchers, marketers, human-resources professionals, professors and students — anyone whose work generates recordings faster than they can review them. The company says its users include professionals at Google, Apple, Amazon, Meta, Microsoft and McKinsey, though these are individual users rather than named enterprise contracts.
Geographically, Mergermarket reported in April that about a third of users are in the United States, a third in Europe, and the remainder spread across Southeast Asia and Latin America. That footprint matches the investor list and the Singapore and Hong Kong offices.
Clipto Pricing: What the Pro Plan Costs
Clipto sells one paid tier, Clipto Pro, on top of a free trial. The live pricing page in early September 2026 shows a higher list price than the $8.99 to $9.99 a month Kang described to Mergermarket in April, which suggests the company has been testing what a professional audience will pay.
| Plan | Price | Notes |
|---|---|---|
| Pro, monthly | $24.99 per month | New-customer discount of $9.99 for the first month |
| Pro, yearly | $12.49 per month, billed yearly | Advertised as 50% off; about $149.88 a year |
| Free trial | 7 days | Desktop app requires Apple Silicon with 16GB, or Windows with 12GB RAM |
| Pro features | Included | Unlimited video and audio search, speech transcription, people tagging, 99+ languages, videos up to 6 hours, advanced models |
The arithmetic favours annual commitment heavily: twelve months at $24.99 comes to $299.88, against $149.88 on the yearly plan. The chart compares the monthly cost under each option.
For context, “hundreds of thousands” of subscribers at the yearly rate of $149.88 would generate tens of millions of dollars a year, which is consistent with the roughly $15 million ARR figure once a mix of discounted, monthly and lapsed subscriptions is taken into account.
How Clipto Compares With Adobe, Apple, Google and TwelveLabs
TechCrunch framed the competitive picture directly: Adobe already offers AI-powered search inside Premiere, and both Apple Photos and Google Photos let users find photos and videos with natural-language descriptions. Kang’s rebuttal is about scope. Clipto searches across videos, audio, images and documents wherever they sit on the machine, while the platform vendors “generally focus on files stored within their own services”.
| Product | Approach | Where it runs | Money behind it |
|---|---|---|---|
| Clipto | Personal media search across video, audio, images and documents; MCP connector for AI agents | On-device (desktop); cloud on mobile | $15M at a $250M valuation, August 2026 |
| Adobe Premiere and Frame.io | AI search inside the editing suite; cloud review and collaboration | Desktop and Adobe cloud | Frame.io acquired for $1.275B in August 2021 |
| Apple Photos and Google Photos | Natural-language search of photos and videos in the vendor’s own library | Device plus vendor cloud | Bundled with the platform |
| TwelveLabs | Video-understanding foundation models (Marengo, Pegasus) sold as an API to enterprises | Cloud API | $100M Series B co-led by NEA and NAVER Ventures; $30M in December 2024 |
| iconik (Backlight) | Cloud media asset management with AI tagging for broadcasters and brands | Cloud | Acquired in Backlight’s $200M PSG-backed launch, April 2022 |
| Google Cloud Video Intelligence | Label, object, shot, speech, text and face detection as an API | Cloud API | Part of Google Cloud |
| Limitless (formerly Rewind) | Personal AI memory from screen and audio capture | Device plus cloud | $33M raised; acquired by Meta in December 2025, Rewind discontinued |
The video-understanding money trail
The comparison table shows how much capital has already flowed into making video searchable. TwelveLabs, the closest pure-play in video understanding, raised a $100 million Series B co-led by NEA and NAVER Ventures with participation from Amazon, Radical Ventures, Index Ventures and others, on top of $30 million in December 2024 from Databricks, Snowflake Ventures and SK Telecom. Adobe paid $1.275 billion for Frame.io in 2021. Backlight assembled iconik, ftrack and three other companies in a $200 million launch in 2022. Meta bought Limitless, the personal-memory startup that had raised $33 million from a16z, NEA and Sam Altman, in December 2025.
Against that backdrop, Clipto’s $15 million is small money — but it is attached to a profitable business, which none of the others could claim at the equivalent stage.
Where Clipto’s approach differs
Three distinctions matter. First, Clipto is personal and cross-format: it indexes whatever is on the disk, not what lives inside one vendor’s library. Second, it is on-device by default, whereas TwelveLabs, iconik and Google Cloud are cloud APIs that require uploading the footage. Third, it is priced as a consumer subscription rather than as enterprise infrastructure, which is how a 20-person company reaches 30 million users. The trade-off is hardware: on-device video understanding needs a modern machine with plenty of memory.
What Clipto's "Memory Layer" Means for Businesses
Kang’s most quotable line from the announcement is the strategic one: “We believe AI needs a memory layer, not just a reasoning layer. Video is where that challenge is most visible today, but the broader opportunity is helping AI understand the information people have accumulated over years of work.”
For most organisations that accumulated information is dark data — recorded meetings nobody rewatches, training videos, site-survey footage, webinar archives, customer-call recordings and years of marketing rushes. A tool that makes all of it searchable, and exposes it to AI agents through MCP, turns a storage cost into a working asset. That is the practical case for including media search in an AI strategy alongside the usual document and database projects.
The same logic applies to how firms manage the archive itself. Video is the fastest-growing category in most companies’ data management and analytics estate, and it is almost always the least catalogued. Clipto’s approach — index locally, expose selectively — is a pattern worth copying even for teams that never buy the product.
Practical questions before adopting on-device media search
A few checks belong in any evaluation. Hardware first: the desktop app’s 16GB Apple Silicon requirement rules out older fleet laptops, and the Windows build is still labelled beta. Governance second: MCP access is scoped and user-authorised, but a business deploying autonomous AI agents needs a policy on which agents may query which archives, and an audit trail of what they retrieved.
Compliance third: automatic people tagging in recordings of staff or customers touches biometric and data-protection rules in the UK and EU, so the on-device default helps but does not settle the question. Cybersecurity fourth: a searchable index of every recording on a laptop is exactly the file an attacker would want, which makes device encryption and endpoint controls part of the rollout rather than an afterthought. Finally, portability: check how notes, transcripts and tags can be exported if the vendor is acquired, as Rewind’s users discovered when Meta bought Limitless.
The risks in the Clipto story
The bull case is clear, but there are honest caveats. No lead investor was named, and the earlier funding history is inconsistent across sources. The list price has more than doubled since April, and consumer subscriptions are sensitive to that. Apple, Google and Adobe can all bundle better search into products people already pay for. And a 20-person team is thin for a company promising native apps on four platforms plus a growing set of agent integrations.
Set against those risks are the numbers that actually got the round done: real revenue, real profit, two-year retention, and a customer base that widened from creators to professions of every kind without a sales team. For a business weighing custom software development for its own media archive, Clipto is also a useful benchmark for what a small, focused team can build on top of today’s on-device models.
Frequently Asked Questions
What is Clipto?
Clipto is an AI-powered media search application founded in 2023 by Henry Kang and headquartered in San Francisco. It indexes videos, audio, images, meeting recordings and documents on a user’s computer so they can be searched in plain language, and it exposes that index to AI assistants such as ChatGPT and Claude through the Model Context Protocol.
How much is Clipto worth?
Clipto is valued at $250 million post-money following a $15 million all-equity round announced on 31 August 2026. TechCrunch reported annual recurring revenue of roughly $15 million at the start of 2026 and profitability on a net-income basis.
Who are Clipto’s investors?
The round was backed by HSG (formerly Sequoia China), GL Ventures, EnvisionX Capital, Palm Drive Capital and 522 Ventures, together with the individual investors Hans Tung and Lu Zhang. No lead investor was disclosed.
Does Clipto send my videos to the cloud?
Not on desktop. The Mac and Windows apps process and index media locally, and the company says users can choose to keep all search on-device. The iPhone and Android apps are cloud-powered, and any AI assistant connected through MCP only receives what the user explicitly authorises.
How much does Clipto cost?
Clipto Pro lists at $24.99 a month, or $12.49 a month when billed yearly, with a $9.99 first-month offer for new customers and a seven-day free trial. Kang had described pricing of $8.99 to $9.99 a month to Mergermarket in April 2026, so the price has risen during the year.
References
Clipto uses AI to search terabytes of video and is now valued at $250M — TechCrunch
Clipto Raises $15M to Build the Memory Layer for the AI Era — company announcement
Clipto.ai plans USD 50m–USD 100m capital raise within six months — Mergermarket
Clipto uses AI to search terabytes of video — Yahoo Finance syndication
TwelveLabs raises $100M Series B
Twelve Labs secures $30 million in funding
Adobe announces intent to acquire Frame.io — Frame.io blog
iconik announces its acquisition by Backlight
Meta acquires Sam Altman-backed AI wearable startup Limitless — SiliconANGLE
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.