Skills Based Upscaling Text automatically translated in your browsing language Automatically translated

Reason for exception Article 2.32(1b) Technical/artistic reasons or protection of exclusive rights I consider that the exception ground for technical exclusivity within the AW 2012 and therefore the choice of one specific supplier is justified for the following reasons: Skill based consists of 6 phases. At the beginning of 2024, this project - Skill based Scaling- started with the aim of developing an AI solution for Defence. Phases 1 and 2 have been completed and completed via the Framework Agreement for Regular Hire. The present request concerns phases 3 to 6 and the nature of the service differs substantially: it is a concrete final result, which builds directly on the previous phases. This solution has been and will be developed specifically for Defence and is therefore not comparable to standard solutions (tools) available on the market. A generic tool purchased off-the-shelf does not meet Defense's unique requirements in terms of safety, process design and integration. Below is an explanation of why this solution should be developed specifically for Defence and why a standard solution is insufficient: 1. The solution runs within Mulan and must integrate seamlessly with existing, often confidential IT infrastructure. This requires customization that generic tools do not offer. 2. Defense works with sensitive information. A standard tool does not provide sufficient guarantees for data security, logging, access control and airgapping required in military contexts. 3. The models are trained on Defense-specific data (such as mobilization and preparation and military tasks). Generic tools are not trained in this context and do not sufficiently understand military terminology or scenarios. 4. The Ministry of Defence is the owner of intellectual property (IP). Only with a custom solution can full control over source code, training data and model output be guaranteed. 5. An off-the-shelf tool often uses third-party infrastructure (e.g. cloud AI services), which increases the risk of data exfiltration or influence on modelling behaviour by external parties. 6. The solution includes custom guardrails and security to ensure the quality of the AI output. This is only achievable with a solution designed from the ground up for Defence. 7. The interface and functions are tailored to readiness and mobilization (Chapter 1). A generic tool does not support this unique process. 8. Within this solution, specific prompts are written and iteratively improved based on internal Defense data. The party that has built up specific tailor-made expertise in the first two phases, also within the context of Defence, is Faraday. Continuation with this party is the most responsible choice in terms of content and organization because of the technical and content-related coherence, the accumulated knowledge and the need for continuity. In fact, there are 2 deadlines, 1 January 2027 it must be operational, for this the model June 2026 must be technically ready. In a time frame in which acceleration has been made an important main objective, a different acquisition choice, regardless of the substantive argumentation, would completely obscure the already under pressure deadline for the (indirect) main task 1-related implementation of this project. For this reason, there is a separate contract with an obligation to achieve a result and this justifies a targeted award to the current supplier, in accordance with Article 2.32(1)(b) of the Aw 2012. Text automatically translated in your browsing language Automatically translated

72000000 - IT services: consulting, software development, Internet and support