Israeli defense-AI startup Airis Labs emerged from stealth and announced it has raised $60 million in total funding, including a $31 million Series B led by PSG Equity (Globes, The Jerusalem Post, The Next Web). Founded in 2023, the Tel Aviv-based company says its video-first platform ingests fragmented visual sources, smartphones, social media, drones, CCTV, and body cameras, and converts them into machine-readable intelligence it calls "User-Generated Field Intelligence" (Globes, The Next Web). Cofounders are Noam Friedman (CEO), Rotem Abeles (CPO) and Us Amos Lahav (GM) (Globes, The Jerusalem Post). Friedman is quoted describing the product proposition as solving a shortage of "machine-readable understanding" for government teams (Globes). Reporting names participation from TLV Partners, Stepstone Group, Redseed Ventures and angel backers including Eyal Waldman, Jeff Horing, Yasmin Lukatz, and David Chinn (Globes, The Next Web). Dealroom and other outlets report deployments with government organisations worldwide and customers including the US Army (Dealroom).
What happened
Airis Labs, an Israeli defence-AI startup, emerged from stealth and announced it has raised $60 million in total funding, including a $31 million Series B led by PSG Equity (Globes, The Jerusalem Post, The Next Web). Reporting identifies participation from TLV Partners, Stepstone Group, Redseed Ventures, and angel investors Eyal Waldman, Jeff Horing, Yasmin Lukatz, and David Chinn (Globes, The Next Web). The company was founded in 2023 and lists cofounders Noam Friedman (CEO), Rotem Abeles (CPO) and Us Amos Lahav (GM) (Globes, The Jerusalem Post). Dealroom and Dealroom-syndicated coverage report deployments with government organisations worldwide and name the US Army among users (Dealroom).
Technical/product details
Airis describes its offering as a video-first intelligence platform that ingests fragmented visual inputs, smartphone uploads, social-media imagery, digital forensics, CCTV, body cameras, and drone footage, and produces structured, machine-readable outputs the company calls "User-Generated Field Intelligence" (The Next Web, Globes). The platform is framed as combining large-scale ingestion, automated structuring, and downstream queryable intelligence for analysts and AI agents; company statements emphasise real-time operational use and agent-scale automation (Globes, The Next Web).
Editorial analysis - technical context
Companies building tooling for heterogeneous visual data typically need robust ingestion, multimodal representation, temporal tracking, and scalable indexing. Industry-pattern observations: systems intended for operational intelligence generally combine object and action detection, multi-source alignment, temporal stitching, and metadata extraction (video timestamps, geo-tags) before higher-level entity and event reasoning. For practitioners, the most time-consuming work in such stacks is reliable data curation and cross-source alignment under noisy, partial, and adversarial field conditions.
Market and competitive context
Industry reporting places Airis within an active defence-AI cluster that includes established vendors and several startups. The Next Web mentions competitors and adjacent offerings from firms such as Anduril and the defence-oriented activity from data-labeling and perception vendors. The move to emphasise "user-generated" field footage attempts to delineate a product niche distinct from classic video analytics, open-source-intelligence platforms, or generic data-fusion tools (The Next Web, Globes).
Industry context
Editorial analysis: Funding at the Series B scale led by a US growth-equity firm signals investor appetite for solutions that can operationalize visual data at scale in government and defence settings. Observed patterns in similar late-seed to Series B rounds show follow-on capital is often used to build out sales and integrations with government procurement pathways, hire domain engineering teams, and scale cloud/edge deployments. Reporting specifically states the new funds will be used to expand US operations (Globes). TechFundingNews reports the company had about 45 employees and is aiming to double headcount by year-end; that headcount and hiring target are company-reported in coverage and should be read as the startup's expectation as presented to press (TechFundingNews).
What to watch
Observers should track the following indicators over the coming 6-12 months:
- •announced contracts or procurement vehicles in the US or allied governments
- •integrations with defence ecosystems such as Oracle Defense or other platform partnerships (The Next Web notes selection into the Oracle Defense Ecosystem)
- •evidence of field performance metrics or red-team evaluations under adversarial conditions
- •product support for edge or on-prem deployments important for classified environments
Limitations on reporting and company commentary
What is reported in press coverage are funding amounts, investor names, cofounder identities, and customer claims. The company has provided quoted product framing through CEO Noam Friedman in coverage; beyond that, public reporting does not include detailed architecture diagrams, benchmark numbers, or independent performance audits in open sources. Industry readers should treat deployment claims as company-reported unless independent verification appears in procurement notices or third-party evaluations.
Takeaway for practitioners
Editorial analysis: The story is notable for practitioners working on multimodal perception and operational AI because it highlights a commercially funded attempt to convert unstructured, user-generated visual material into structured, actionable intelligence at scale. Key technical challenges to follow include robust temporal stitching, multi-source geolocation, and handling low-quality user-generated imagery under real-world adversarial conditions.
Scoring Rationale #
A mid-size Series B for a defense-AI startup is notable for practitioners tracking operational video intelligence and procurement in government markets. The story matters for teams building multimodal ingestion and indexing but is not a frontier research breakthrough.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.