Military simulation procurement fails on three questions more often than on price: which of the three simulation types is actually being sold, whether the product will interoperate with the simulations the customer already owns, and whether anyone has accredited it as fit for the specific training purpose. Each question has a published standard behind it. None of the three is answered by a demonstration.
Simulation and training is one of the denser categories in the Israeli Defence-AI Deployment Index, with five companies. Two carry a public report of use by a military; three carry an announced contract. The category page sets out each entry. What follows is the procurement vocabulary that sits underneath those entries.
Live, virtual and constructive are three different products
The US Department of Defense modelling and simulation glossary, maintained by the Modeling and Simulation Coordination Office and reproduced in service-level glossaries, defines the taxonomy precisely. A live simulation involves real people operating real systems. A virtual simulation involves real people operating simulated systems, which is where flight, vehicle and weapon trainers sit. A constructive simulation involves simulated people operating simulated systems, with real people making inputs but not determining outcomes.
The distinction is not academic. Live simulation requires instrumentation, range infrastructure and after-action review tooling. Virtual simulation requires fidelity in the human-machine interface and, increasingly, in the rendered environment. Constructive simulation requires validated models of behaviour, which is the hardest of the three to evidence and the one where AI claims cluster. A vendor selling one and being evaluated against the requirements of another is the most common source of a failed acceptance test.
Vendors frequently offer a blend. Where they do, ask which component is live, which is virtual and which is constructive, and require the answer in the vendor's own written response rather than in a slide.
Interoperability standards decide the lock-in
Distributed simulation only works if federated systems speak a common protocol. The two long-standing families are Distributed Interactive Simulation and the High Level Architecture, the latter standardised by the IEEE in the 1516 series. NATO has ratified HLA under Standardization Agreement 4603 and Command and Control Systems to Simulation Systems Interoperation under STANAG 4856, and the NATO Modelling and Simulation Group publishes the master list of approved standards as the NATO Modelling and Simulation Standards Profile, AMSP-01.
Two practical points follow. First, federations built on different architectures are not inherently interoperable with one another, so a customer running an HLA estate should treat a DIS-only product as a gateway project, not a plug-in. Second, standards compliance is a documented state, not an assertion. Ask for the specific edition, the federation object model, and the record of any interoperability event the product has taken part in.
The NATO Modelling and Simulation Master Plan, approved by the North Atlantic Council, is the governing policy document for alliance M&S coordination and is worth reading before writing a requirement that will need to sit inside a coalition exercise.
Accreditation is the question most tenders skip
US policy on this is explicit. DoD Instruction 5000.61 establishes the framework for verification, validation and accreditation of models, simulations and associated data, and the version effective from September 2024 requires that documentation identify the context, version and intended use of the model being accredited. That last clause is the one buyers should hold on to: accreditation is granted for a stated purpose, not in general.
Verification asks whether the model was built as specified. Validation asks whether it represents the real world well enough for the intended use. Accreditation is the customer's formal decision that it may be used for that purpose. A vendor cannot accredit its own product. If a proposal describes a simulation as "validated" without naming the accrediting authority, the intended use and the date, the word is doing no work.
For AI-driven components the question is sharper still. A behaviour model that generates opposing forces from learned data needs a stated basis for its behaviours, and a customer's validation authority needs to be able to interrogate it. Ask what happens when the model is asked to represent a situation absent from its training data.
What the Israeli simulation and training entries show
The index grades on published evidence only, under the four published grades.
Two entries carry reported use. Israel Defense reported on Bagira Systems, which builds instrumented live training systems. NoCamels reported on Combatica's virtual-reality platform used to train security forces. Both are third-party reports rather than company announcements, which is what places them in the operational band.
Three entries carry announced contracts without a published account of use. The Times of Israel reported the Israeli Ministry of Defense selecting Cognata for autonomous driving simulation work with the IDF. NoCamels reported Cympire, a cyber-range company, contracted to provide training for the IDF. Shephard Media reported Simlat awarded a contract for advanced unmanned aircraft system training.
That distribution is typical of the category and is worth reading carefully. Training procurement is announced more readily than most defence contracting, because there is rarely an operational-security reason to withhold it. The result is that simulation companies tend to have better public paper than their peers in other categories, which is a fact about disclosure practice rather than about product quality.
Five things to establish before contract
Ask which simulation type is being supplied, in the taxonomy above. Ask for the interoperability standard and edition, with evidence of a federation event. Ask who accredited the model, for what use, and when. Ask what the after-action review data model is and whether the customer owns the exercise data. Ask what the content pipeline costs: scenario authoring, terrain and model updates are where lifetime cost accumulates, and they are rarely in the headline price.
The general method for testing a supplier's public record is set out in the verification guide, the companies with documented operational use are listed in full here, and the reason a high readiness level is not the same as evidence of use is covered in the readiness-level guide. Where autonomy is being simulated as well as trained, the autonomy due-diligence checklist applies in parallel.
Limits of this reading
The standards cited are US and NATO. National simulation policies differ, and a buyer outside those frameworks should check its own accreditation regime rather than assume the DoD instruction applies. Standards editions change; the numbers here are current as published at the date of this article and should be verified against the issuing body.
The index entries described above record what was published, by whom and when. They establish nothing about training effectiveness, transfer of learning, or fidelity. No public source in the index measures those things, and the index does not attempt to.
Frequently asked questions
What is the difference between live, virtual and constructive simulation?
Live simulation involves real people operating real systems. Virtual simulation involves real people operating simulated systems, such as a vehicle trainer. Constructive simulation involves simulated people operating simulated systems, with real people providing inputs but not determining outcomes.
Which interoperability standards matter in defence simulation?
Distributed Interactive Simulation and the IEEE 1516 High Level Architecture are the two main families. NATO has ratified HLA under STANAG 4603 and lists approved standards in the NATO Modelling and Simulation Standards Profile, AMSP-01. Different architectures are not interoperable without a gateway.
What does accreditation mean for a military simulation?
Accreditation is the customer's formal decision that a model or simulation may be used for a specific stated purpose. US policy in DoD Instruction 5000.61 requires documentation identifying the context, version and intended use. A supplier cannot accredit its own product.
Sources
- US Department of Defense, Instruction 5000.61, DoD Modeling and Simulation Verification, Validation, and Accreditation
- US Army, Modeling and Simulation Glossary
- NATO Science and Technology Organization, NATO M&S Standards Profile (AMSP-01)
- NATO Science and Technology Organization, NATO M&S Master Plan
- Israel Defense, report on Bagira Systems
- NoCamels, VR platform trains security forces in hyper-real scenarios, March 2024
- The Times of Israel, Defense Ministry taps Israeli firm for autonomous driving simulation with IDF
- Shephard Media, Simlat awarded contract for advanced UAS training
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.