Counter-UAS produced eleven entries in the Israeli Defence-AI Deployment Index. Three other categories produced one apiece. Active protection, targeting and fire control, and military logistics and sustainment each yielded exactly one company that met the inclusion test, and the emptiness of those Israeli defence AI categories is reported here as a finding rather than a research shortfall.
The single entries are BIRD Aerosystems in active protection, graded L2; SMARTSHOOTER in targeting and fire control, graded L3; and Odysight.ai in logistics and sustainment, graded L2. There is no second name in any of the three.
What the three thin categories contain
BIRD Aerosystems, founded 2001 in Herzliya, appears on the strength of European Security & Defence reporting on delivery of its airborne missile protection system. It is the whole of Israeli active protection AI.
SMARTSHOOTER, founded 2011 at Kibbutz Yagur, appears on Breaking Defense reporting that the US Army is expanding use of its fire-control system. It is the whole of targeting and fire control AI, and the only one of the three thin categories with an entry at the operational grade under the index grading rules.
Odysight.ai, founded 2013 in Omer, appears on a Nasdaq press release announcing a purchase order from an unnamed international defence contractor. It is the whole of military logistics and sustainment AI.
By contrast, counter-UAS carries eleven entries spread across every grade, and ISR and sensor fusion, autonomous air systems and simulation and training each carry five or more. The distribution across the 50 entries is lopsided in a way that repays explanation.
Active protection is a primes' category
Vehicle active protection in Israel is not a startup field. Defense Update's survey of the sector describes Rafael's Trophy as the most widely fielded system, adopted for US Abrams tanks, British Challenger 3 and German Leopard 2A8, with European work handled through the EuroTrophy consortium. Elbit Systems produces the competing Iron Fist family, originally developed by Israeli Military Industries, and Nasdaq reported Elbit receiving further US Army contracts for it.
Both are heavy-integration products. They require ballistic testing ranges, vehicle integration authority and a relationship with a platform manufacturer, and they are sold as part of a vehicle programme rather than as software. A three-person AI company cannot enter that market, and the index excludes the large primes by design, for reasons set out in why the Israeli primes are excluded.
The result is a category of one. It is a finding about where Israeli startups have not gone, not a claim that Israel lacks active-protection capability. The opposite is closer to the published record.
Fire control has one entry and it is at the top
Targeting and fire control is thin for a different reason. The barrier is not integration weight but regulation: the product attaches to a weapon, which places it inside export-control and end-use regimes that most software companies avoid entirely.
SMARTSHOOTER is the exception that reached the index, and it reached the operational grade. The public sourcing is unusually direct for this sector, with Breaking Defense reporting an expansion of use by a named armed service. One qualifying company at L3 tells a reader more than five companies at L0 would, which is the argument the index is built on and is developed in the deployment evidence gap.
Sustainment is the thinnest of the three
Military logistics and predictive maintenance is a live procurement area internationally. Defense One has argued the future of military logistics is predictive; the US Army has published on predictive logistics as a sustainment direction; DefenseScoop reported the Marine Corps prototyping AI tools for aviation supply and predictive maintenance. Demand exists and is documented.
Israeli startups are barely visible in it. One structural reason is who buys sustainment AI: it is procured by platform sustainment programmes, which contract with large integrators. The UK's Project THESEUS, which examines applying autonomy to battlefield resupply, illustrates the pattern. Trade reporting records Marlborough Communications Limited with Israel Aerospace Industries awarded a demonstration contract, and Rafael separately selected to demonstrate AI-based resupply. Israeli participation is real and it arrives through primes.
A second reason is commercial. Sustainment AI requires access to years of maintenance telemetry from a fleet the vendor does not own. That is a data-access problem before it is an algorithm problem, and it favours incumbents.
Odysight.ai's route in is instructive: a visual-inspection technology adapted from a medical-device origin, sold to a defence contractor rather than directly to a ministry. Even the one qualifying company reaches the category sideways.
What would change the count
Each thin category has a specific unlock. Active protection would need a startup to be integrated into a named vehicle programme and for that integration to be reported. Fire control would need a second company to publish an export licence or a named service customer. Sustainment would need a fleet operator to publish a data-sharing arrangement.
None of those are unlikely over a two-year horizon. The index is versioned for exactly this reason: a category of one is a statement about August 2026, not a structural claim about the sector.
Limits of this reading
The inclusion test drives these counts. The index covers companies whose core product is AI or autonomy applied to defence, and it excludes the large primes. A different boundary would produce different numbers, and readers comparing this index to landscape maps counting between 60 and 334 companies should expect the categories to disagree.
The counts also inherit the sample's bias toward English-language coverage. A sustainment or active-protection startup covered only in Hebrew would not surface here at all, so the thin categories are the ones where an undercount is most likely and least detectable. Thin is what the public record shows; it is not a measurement of the underlying population.
Frequently asked questions
Which Israeli defence AI categories have the fewest companies?
Active protection, targeting and fire control, and military logistics and sustainment each yielded exactly one qualifying company in the index: BIRD Aerosystems, SMARTSHOOTER and Odysight.ai respectively. Counter-UAS, by comparison, yielded eleven entries.
Does a thin category mean Israel lacks capability in that area?
No. Active protection in Israel is dominated by large primes such as Rafael and Elbit, which the index excludes by design. A category of one records where AI-first startups have not entered, not where national capability is missing.
Why are so few Israeli startups in military predictive maintenance?
Sustainment AI is procured through platform programmes that contract with large integrators, and it requires access to fleet maintenance data the vendor does not own. Israeli participation in programmes such as the UK's Project THESEUS has come through primes rather than startups.
Sources
- idf.ai, The Israeli Defense-AI Deployment Index v1.0, 24 August 2026
- European Security & Defence, BIRD Aerosystems AMPS delivery, October 2025
- Breaking Defense, Israel's SMARTSHOOTER says US Army expanding use of high-tech fire control system, June 2025
- Nasdaq, Odysight.ai announces purchase order from an international defense contractor
- Defense Update, Active Protection Systems: an evolving shield, May 2025
- Nasdaq, Elbit Systems gets contracts for Iron Fist active protection system
- Asia Pacific Defence Reporter, MCL and IAI awarded UK MoD Project Theseus contract
- DefenseScoop, Marine Corps prototyping AI tools for aviation supply, predictive maintenance, April 2026
Independent publication of idf.ai. Not affiliated with, endorsed by, or connected to the Israel Defense Forces, the Israeli Ministry of Defense, or any government body. Compiled entirely from publicly published sources. No classified, restricted or non-public information. Listed companies may dispute any entry: send the published source that contradicts it and the entry will be amended or removed.