MMoments Lab

Moments Lab doubled its recurring revenue in a year and raised $24 million in 2025 by selling one idea to newsrooms and studios: every second of their archive should be searchable by what happens in it. Customers include Thomson Reuters, Sinclair, Hearst, Amazon Ads and Banijay. Its MXT-2 model indexes footage by scene, person, action, place and soundbite, and its Discovery Agent lets an editor ask for a moment in plain language and get it back.

It is best for broadcasters, news agencies, studios and sports organizations sitting on thousands of hours of video. It is an enterprise purchase priced by quote, and the agent was launched in beta for existing customers in September 2025. We rate it 7.5/10 on our scale. We have not used Moments Lab ourselves; this review is based on the company’s product pages and announcements and on reporting from French deal site CFNEWS.

Business Figures From French Financial Press

Moments Lab has shared more about its business than most media AI vendors, partly through French financial press coverage of its 2025 round.

Read together, these figures describe a focused enterprise vendor rather than a mass-market tool. About 60 customers and just under €5 million in recurring revenue imply a relatively small number of sizable contracts, and the customer list is weighted toward news agencies, broadcasters and studios with very large archives.

The growth and the 2025 round suggest the company has the runway to keep building the agent. For buyers, the most practical facts are the data commitment, that client footage stays inside the platform, and the language list, which decides whether the index will work for your newsroom.

Gather the Archive, Index With MXT-2, Then Search

Moments Lab works in three layers, according to its product pages: bring the archive together, index it with MXT-2, then search it with the Discovery Agent.

Step One: Centralize the Archive

Teams connect footage from on-premises storage, cloud storage and live feeds. The site lists Amazon S3 and Adobe Premiere Pro among its integrations, so footage can come in from cloud buckets and selected moments can move into an editor.

Step Two: Index With MXT-2

MXT-2 is a patented multimodal model. It splits each video into scenes and identifies people, actions, locations and shot types, and it picks out the best soundbites. Customers can train it to detect specific people or logos, or to describe content in the organization’s own tone. The output is time-coded metadata, which the Discovery Agent needs before it can search anything.

Step Three: Ask the Discovery Agent

• Ask: an editor types a request in plain language, such as a type of scene, a person or a theme.
• Refine: follow-up questions narrow the results, and the agent remembers the earlier context.
• Retrieve: results come back as specific moments with timecodes, not just whole files.
• Suggest: the agent recommends clips and concepts, and can surface footage the team did not know it had.
• Enrich: it can crawl the web and draw on custom data sources for context.

Moments Lab pitches the agent for news pieces, trailers, social posts, recaps, documentaries, brand activations and archive projects, and says non-technical users can work with it. CEO Philippe Petitpont has said the aim is to let teams generate rough cuts in minutes with a simple prompt.

Where It Fits in a Newsroom

In most newsrooms, archive search still runs on whatever metadata someone typed in when the footage was filed: a slug, a date, a location and perhaps a few names. Anything not written down is hard to find. A producer looking for a specific politician shaking hands at a summit, or a wide shot of a stadium at night, often ends up scrubbing through hours of video or asking a librarian who remembers.

Moments Lab replaces that manual layer with machine-made metadata for every scene, which means footage filed years ago with thin notes becomes searchable by what is actually on screen. The Discovery Agent then turns that index into a conversation. A producer cutting an obituary, an anniversary piece or a social recap can ask for the moments, refine the request and pull the timecodes into an editor.

Editors Build the Story and Clear the Rights

The agent finds and proposes footage. Editors still build the story, check rights and licensing on archive material, confirm facts and context, and approve what airs. In newsrooms, that verification step is the job, so the realistic gain is less time spent searching rather than less editorial judgment.

Pricing Through a Demo or the Sales Team

Moments Lab does not publish prices. Its get-started page points buyers to a demo or the sales team.

Reading Deal Size From Revenue and Customer Count

The company’s reported figures give a rough sense of deal size. Annual recurring revenue just under €5 million across about 60 customers works out to an average of roughly €80,000 a year per customer. That is our arithmetic on figures reported by CFNEWS, not a price list, and individual contracts will vary widely.

• A national broadcaster or news agency: with decades of footage and daily live feeds, expect a deal at or above that average, since indexing volume drives the work.
• A sports federation or event organizer: customers such as UIPM, WRC Promoter and UTMB Group show the product also serves sports organizations with narrower archives, which may mean smaller contracts.
• A digital publisher with modest video: a short archive may not justify an enterprise platform; a self-serve editor such as Descript Underlord is cheaper.

Costs beyond the contract include moving or connecting archives, staff time to train custom detections and review results, and rights checks on old footage. Ask how fees scale with indexed hours, live feeds, users and languages.

Moments Lab Compared With WSC Sports, Magnifi and Descript Underlord

Moments Lab is a search-and-discovery tool for archives. WSC Sports and Magnifi automate live highlights, and Descript Underlord edits recorded video. They overlap where teams want to reuse footage quickly.

If the problem is finding the right shot across years of footage, Moments Lab is the closest fit in this group. If the problem is publishing clips during a live game, WSC Sports or Magnifi are built for that. If the footage is a few recorded shows a week, Descript is cheaper and faster to start. More options are in our AI agents for media and AI agents for video guides.

Newsrooms, Broadcasters and Studios With Big Archives

• News agencies and broadcasters: Thomson Reuters, Sinclair and Hearst show the product is built for newsroom archives and daily feeds.
• Studios and producers: production group Banijay is a named customer, and documentaries, recaps and trailers are listed use cases.
• Sports organizations with deep libraries: UIPM, WRC Promoter and UTMB Group are featured in Moments Lab customer stories.
• Brands with large video libraries: Amazon Ads is a named customer, and brand activations are a listed use case.
• Multilingual teams: support for English, Spanish, Arabic, Dutch and French covers many European and Middle Eastern newsrooms.

Small Archives and Live Highlight Needs

• Teams with small archives: the indexing platform makes most sense at thousands of hours.
• Live highlight needs: WSC Sports or Magnifi publish clips during games; Moments Lab is built for search.
• Self-serve buyers: pricing comes by quote, without instant signup.
• Newsrooms working mainly in other languages: check the five supported languages against your content first.

Beta Status, Index Quality and Rights Checks

Beta status. The Discovery Agent was released in beta to existing customers in September 2025. Ask what is generally available now, what is still in testing and how feedback shapes releases.

Index quality. CTO Fred Petitpont has said quality indexing metadata is the foundation of a useful agent. If tagging misses a person or mislabels an event, the agent will miss it too. Test with footage that is hard to label, such as crowd scenes or archival film.

Editorial and rights checks. An agent that surfaces forgotten footage can also surface clips with unclear rights, outdated context or people who should not be shown. Keep rights and standards review in the workflow.

Data handling. The company says all processing happens within its platform and that no client video is shared with third-party LLM tools. Ask for this in the contract, along with where data is hosted and what happens to indexes if you leave.

Pricing clarity. Since prices come by quote, confirm how costs scale with archive size, live feeds and users before you index everything.

The Oxx-Led Round With Orange Ventures

• On June 3, 2025, Moments Lab announced a $24 million funding round led by Oxx, with Orange Ventures, Kadmos, Supernova Invest and Elaia, to fund US expansion and research and development.
• On June 6, 2025, CFNEWS reported details of the round, including a valuation of €75 million to €80 million, about 60 customers and 59 employees.
• On September 9, 2025, Moments Lab launched its Discovery Agent in beta for existing users and joined an IBC Accelerator project on AI agents for live production with ITN, the BBC and Channel 4.

Among the Credible Picks for Archives Too Big to Search by Hand

For a newsroom, broadcaster or studio whose archive is too big to search by hand, Moments Lab is one of the more credible options, with a customer list that includes Thomson Reuters, Sinclair and Hearst. It scores 7.5/10 on our scale: solid on autonomy and data handling, with quote-based pricing and an enterprise setup that starts with indexing the whole library as the main considerations.

Its value depends on how often your teams need old footage. If producers spend hours each week hunting for shots, a moment-level index and a conversational search agent can pay off quickly. Ask for a pilot on a slice of your archive, measure how often the agent finds the right moment on the first try, and get pricing for full-scale indexing before signing.

MXT-2 and Discovery Agent Access for New Clients

How Much Does Moments Lab Cost?

Moments Lab does not publish prices; buyers book a demo for a quote. Reported figures of just under €5 million in recurring revenue across about 60 customers suggest enterprise-sized contracts.

What Is MXT-2?

MXT-2 is Moments Lab’s patented multimodal indexing model. It splits videos into scenes and tags people, actions, locations, shot types and the best soundbites, creating the time-coded metadata the Discovery Agent searches.

Is the Discovery Agent Available to New Customers?

It launched in September 2025 as a beta for existing users. New customers get it as part of the platform, and it only works on footage indexed by MXT-2.

Does Moments Lab Use Client Video to Train Outside AI Models?

The company says all processing happens securely within its platform and that no client video content is shared with third-party LLM tools. Buyers should confirm data terms in their contract.

Was Moments Lab Called Newsbridge?

Yes. The company was founded in 2016 by twin brothers Philippe and Fred Petitpont as Newsbridge and changed its name to Moments Lab in 2024.