AI Challenge Competition

The AI-BOOST AI Challenge Competition is an EU-funded open innovation call for generative AI solutions running multiple technical tracks, including Challenge 1 (GenAI-based natural language mission generator for autonomous agricultural robots) and Challenge 2 (agentic AI for automated CAD generation and autonomous simulation). Eligible applicants are individuals, teams, SMEs, startups, research organisations and consortia established in EU Member States or eligible Associated Countries, with demonstrated AI technical capacity and a focus on civil applications. The competition uses a two-phase process (SPARK concept-note selection and ADVANCE five-month development) with Spark prizes of €28,500 for five winners per challenge and a €100,000 grand prize for one final winner per challenge, plus special awards; concept notes for Challenges 1 and 2 must be submitted via the F6S portal by 25 August 2026, 17:00 CEST. Selected teams receive technical mentoring, access to challenge datasets and optional HPC resources, and finalists will present live at the February 2027 final event in Brussels.

What it funds

A multi-topic GenAI challenge programme (AI-BOOST) running four technical tracks. It funds rapid development and validation (proof-of-concept TRL 4–6) of generative AI solutions in autonomous robotics, agentic CAD and simulation, clinical dataset synthesis, and automated test-case generation for mobility.

Who can apply

Open to individuals and teams from academia and industry (students 18+, researchers, developers, start-ups, SMEs, NGOs, consortia). Applicants must have technical AI capacity, focus on civil applications, and be established in EU Member States, certain overseas territories or eligible Associated Countries.

Key deadlines:Challenges 1 and 2: 25 August 2026, 17:00 CEST. Challenges 3 and 4: 8 September 2026, 17:00 CEST.

Funding and prizes:Spark phase: five winners per challenge receive €28,500 each. Advance phase: one final winner per challenge receives €100,000. Two special awards per challenge: Innovation Excellence €12,500 and Responsible AI €12,500.

1. 1Two-phase competition: Spark (concept note selection) then Advance (five-month development, mid-term checks, final live demo).
2. 2Evaluation covers technical excellence, innovation, scalability, responsible AI and real-world applicability.
3. 3Participants receive mentoring, challenge-specific data and optional HPC access (CINECA LEONARDO).

How to apply

Applications (concept notes) must be submitted via the F6S platform following registration and completion of the application form. Full submission guidance and challenge descriptions are on the AI-BOOST site and F6S listing AI-BOOST competition page. ¹

1. 1Register on F6S and create your user profile.
2. 2Complete the mandatory fields in the concept note form.
3. 3Submit before the relevant challenge deadline and follow guidance for required supporting documents.

Total programme budget indicated on the call page:€535,000 (funded under Horizon CL4 / AI-BOOST GA No 101135737).

Footnotes

1. 1Submit applications via the F6S portal as directed on the AI-BOOST website: f6s.com (support and platform instructions available there).

Basic Facts

Call identity and context

Title:AI Challenge Competition (part of AI-BOOST). Call identifier: HORIZON-CL4. Project acronym: AI-BOOST. Funded under the European Union Grant Agreement No 101135737. Competition published as an Open innovation activity under the AI Data and Robotics Partnership (CSA) to accelerate generative AI research and adoption across industrial and domain-specific use cases. Relevant portal reference: European Commission Funding & Tenders Portal listing for the competitive call. The AI-BOOST competition targets Generative AI domains, launching four challenge tracks led by industry and research owners.

Deadlines:Challenges 1 and 2: 25 August 2026, 17:00 CEST. Challenges 3 and 4: 8 September 2026, 17:00 CEST.

Duration and budget:Competition duration: two sequential phases spanning approximately eight months at programme level; the Advance Phase development period for winners is five months. Total budget indicated for the specific call: €535,000 (call-level figure).

Challenges and technical focus

The AI Challenge Competition runs four distinct challenge tracks. The scraped content includes full published descriptions for Challenges 1 and 2 and summary descriptions for Challenges 3 and 4. Each challenge is led by an industry or research Challenge Owner who provides datasets, infrastructure access, mentoring, and evaluation criteria. The competition seeks solutions that start at TRL 2-3 and reach TRL 4-5, producing modular proofs of concept demonstrated in representative environments.

1. 1Challenge 1: GenAI-Based Natural Language Mission Generator for Autonomous Robots in Agriculture — led by Consorzio Intellimech and JOiINT LAB (Italy). Technical focus: Vision-Language-Action models to convert spoken and multimodal inputs into ROS-compatible mission plans. Target TRL: 5. Use case: vineyard productivity; extendable to inspection and maintenance.
2. 2Challenge 2: Agentic AI for Automated CAD Generation and Autonomous Simulation — led by SIAD Group (Italy). Technical focus: agentic generative AI to convert natural language and 2D/3D CAD inputs into parametric CAD models, automatic mesh generation, convergence analysis, and simulation-ready outputs for piping and skid designs. Target TRL: 5-6.
3. 3Challenge 3: Generative AI for Enhancement of Clinical Datasets — led by EUCAIM. Technical focus: synthetic clinical imaging data generation to fill gaps and reduce bias; harmonisation across imaging conditions.
4. 4Challenge 4: Generative AI for Automatic Test Case Generation from Crash Databases & Standards — led by Siemens Industry Software NV (with EU RobustifAI). Technical focus: automatic scenario generation for autonomous driving safety simulations from accident reports, visual data and standards.

Eligibility and who should apply

Participation is open to individuals, teams, and legal entities from academia and industry. Applicants can be natural persons linked to a legal entity or apply as individuals. The competition explicitly targets multidisciplinary teams with AI and domain expertise suitable for civil applications only.

1. 1Eligible Applicant Types: Individuals (students or researchers above 18), startups, SMEs, high-tech companies, spin-offs, universities, research institutes, NGOs, non-profit organisations, or consortia/teams composed of these. Applicants must demonstrate technical capacity and experience working on AI systems.
2. 2Geographic Eligibility: Applicants must be established in one of the EU Member States (including overseas countries and territories) and some Associated Countries. Applicants should consult the guide for applicants for the exact list of eligible Associated Countries. The programme is funded by the EU and primarily addresses EU-based applicants and associated countries.
3. 3Application Mode: Open call via F6S platform. Applicants must register on F6S, complete mandatory application fields, upload supporting documents, and submit before the stated deadline. Non-electronic submissions are not accepted.

Funding model, prizes and nature of support

This is a prize-based competition with non-repayable cash awards and additional in-kind and non-monetary support. The call uses a grant/award-style mechanism embedded within an open innovation competition structure. Financial prizes are disbursed to winners at the end of phases; additional technical, mentoring, and infrastructure support is provided throughout the Advance Phase.

Nature of Support:Money and non-monetary services: winners receive cash prizes and also technical guidance, mentoring, access to challenge-specific datasets, optional HPC access (CINECA LEONARDO), onboarding and training, and opportunities for collaboration with Challenge Owners and dissemination support.

Application and evaluation process

Applications are submitted through the F6S platform. The evaluation is performed in successive phases: Spark Phase evaluation of concept notes and Advance Phase continual monitoring, mid-term checkpoint, final report, final algorithm submission, final pitch, live demonstration and audience voting. External experts will assess applications during Spark Phase; evaluators are remunerated per application reviewed.

1. 1Registration and application via F6S platform: create user profile, complete mandatory fields, upload supporting documents and submit before deadline.
2. 2Spark Phase evaluation: concept note assessed for novelty, technical feasibility, alignment to challenge objectives; five winners per challenge selected.
3. 3Advance Phase development: five-month programme including technical mentoring, validation, access to data/infrastructure, a mid-term checkpoint to verify progress.
4. 4Final evaluation and event: final report, final algorithm submission, a live demo and pitch at the Final Live Event (Brussels, February 2027). One winner per challenge selected.

Application Stages:Two principal stages: Spark Phase (concept note) and Advance Phase (development and demonstration). Evaluation within each stage includes internal and external reviews, mid-term checkpoint (during Advance), and final live assessment.

Application Type:Open call via the F6S platform. Applicants must follow guidelines for applicants available on AI-BOOST website and submit through F6S. The call is not invitation-only; it is competitive and open to eligible applicants.

Project maturity, expected outcomes and evaluation metrics

The competition targets early-to-mid technology readiness projects, progressing from TRL 2-3 to TRL 4-5 (Challenge 1 target TRL 5; Challenge 2 target TRL 5-6). Expected outputs are modular proofs of concept, validated in representative or real-world environments, with KPIs and quantitative metrics defined per challenge to assess performance, robustness and practical applicability.

1. 1Project Stage: research & development, validation, demonstration and prototype integration (development to TRL 4-6 depending on challenge).
2. 2Expected Outcomes: functional PoC capable of performing the described tasks (e.g., voice-to-ROS mission generation validated in vineyards, and parametric CAD generation with simulation setup validated in industrial piping scenarios).
3. 3Evaluation Metrics: challenge-specific KPIs. Example from Challenge 1: intent recognition accuracy target ≥ 90%, mission execution success rate target ≥ 85%, robustness across natural language variants ≥ 85%, latency target ≤ 5–10 seconds. Example from Challenge 2: engineering effort reduction ratio (AI hours/human hours) target < 1, package size reduction target < 1 (smaller package vs baseline).

Data, IP, legal and ethical framework

Challenge Owners provide datasets and retain control over data rights according to licensing and nondisclosure agreements. Datasets are a mix of open-source licensed data and proprietary data offered under restricted access with NDAs and cybersecurity policies. Data does not contain personal or sensitive data in the published cases. Participants must respect license terms and owner-specific AI usage policies; data sharing outside authorised contexts is restricted.

Support and infrastructure provided

Challenge Owners (Consorzio Intellimech, JOiINT LAB, SIAD Group, EUCAIM, Siemens) provide mentoring, co-creation, domain expertise and datasets. Optional access to HPC resources (CINECA LEONARDO supercomputer) and cloud/virtual workstation environments may be offered subject to onboarding, training and security compliance. Physical test infrastructures are available where applicable (e.g., Joiintlab vineyard testbed for Challenge 1).

Eligibility checklist and applicant requirements (detailed)

Applicants should carefully verify the following before applying:eligibility by country (EU Member State or eligible Associated Country as per guide for applicants); demonstrable technical capacity in AI and relevant domain (robotics, CAD/CAM/FEA and simulation, medical imaging, or mobility/vehicle safety depending on challenge); compliance with civil-only use case requirement; availability to participate in the five-month Advance Phase and attend the final live event in Brussels (February 2027) if selected; readiness to sign NDAs and comply with data-owner cybersecurity and data-governance requirements where proprietary datasets are provided.

1. 1Applicant must be 18+ for individuals and have legal capacity to accept prizes and contractual terms.
2. 2Teams/consortia must designate a contact/lead and clarify IP and ownership arrangements for developed solutions according to guidelines provided at application.
3. 3Applications must include technical descriptions, concept notes, and evidence of prior work on AI systems. All mandatory fields in the F6S application must be completed.

Consortium requirement

Consortium is optional. Applicants may apply as single entities, individuals affiliated to a legal entity, or as teams/consortia of multiple organisations. The competition accepts single applicants and consortia; legal entity linkage is required for prize receipt and contractual arrangements.

Co-funding and financial obligations

No co-funding from applicants is stated as a requirement for receiving the prize awards; the competition provides fixed cash prizes to winners. However, participants are expected to provide in-kind contributions (time, development effort, possibly computational costs) and to meet technical and validation commitments during the Advance Phase. Any costs associated with using external cloud services or proprietary tools not provided by the programme are the responsibility of participants unless otherwise agreed.

Success rates and selection numbers

Selection is competitive with predefined numbers:per challenge, five winners are selected from the Spark Phase pool to advance; of those five, one final winner is selected in the Advance Phase. Therefore, Spark Phase selection rate depends on number of applicants; guaranteed selection figures: 5 selected per challenge, 1 final winner per challenge. The programme aims to generate 10 breakthrough solutions across included challenges and recognises top candidates for industrial adoption. Exact overall success rate cannot be calculated without total applicant numbers.

Application templates and structure

Applications are submitted through F6S using a structured concept-note form. The guide for applicants and the F6S application form define mandatory sections. While the full F6S form template is available on the platform, the scraped guidance indicates required elements and evaluation-relevant sections:

1. 1Applicant identification and contact details: legal name, country, VAT or company registration, lead contact.
2. 2Team composition and roles: short CVs, relevant experience in AI and domain expertise, institutional affiliation.
3. 3Concept Note / Technical summary: problem statement, proposed approach, novelty, technical feasibility and key innovations. Include planned use of provided datasets and any external datasets or tools.
4. 4TRL and development plan: starting TRL, expected TRL at the end of Advance Phase, workplan for the five-month development period, milestones and deliverables.
5. 5Validation and evaluation plan: how proposed KPIs will be measured against challenge-specific metrics and baseline, planned testing infrastructure (simulation, field, laboratory), safety, explainability and Responsible AI measures.
6. 6Data management and IP: proposed data use, storage, compliance with NDAs and data-owner policies, preliminary IP and exploitation plan.
7. 7Resource and budget estimate: high-level resource needs and justification (computational requirements, personnel time). Note that prize funding is fixed and not a cost-reimbursement grant.
8. 8Ethics and safety statement: risks, mitigation, fail-safes, human oversight arrangements.
9. 9Supporting documents: links to publications, code repositories, previous demonstrations, letters of support or intent (if applicable).

Applicants should consult the official guidelines for applicants and the F6S application page for detailed submission templates and file formats. All mandatory fields must be completed at submission; incomplete forms risk rejection.

Practical considerations and recommendations

Suggested preparatory actions for applicants:review challenge-specific annexes and datasets to align concept notes with provided data and evaluation KPIs; prepare evidence of relevant prior work and reproducible code or demonstration materials; ensure readiness to sign NDAs where proprietary datasets are provided; plan a realistic five-month development roadmap with clear milestones; prepare safety and human oversight strategies for real-world testing; confirm ability to travel to Brussels for the final live event if advancing to the final selection.

Mentioned countries

Explicit country mentions in published challenge documentation:Italy (Consorzio Intellimech, JOiINT LAB and SIAD Group are based in Italy). Geographic scope/eligibility: EU Member States and some Associated Countries; overseas countries and territories of EU Member States are also eligible. No exhaustive Associated Countries list is embedded in the scraped content; applicants should consult the guide for applicants for complete eligibility list.

Templates and evaluation guidance — summary

The application structure requested on F6S mirrors standard concept-note to prototype competition formats. The Spark Phase concept note should be concise but technically specific, addressing novelty, approach, datasets usage, TRL progression, KPIs, and validation strategy. Evaluation will weigh technical excellence, innovation potential, scalability, sector relevance, responsible AI principles, sustainability and practical applicability to the target use case. Advance Phase deliverables include mid-term checkpoint reporting, final algorithm submission, demonstration artifacts and final pitch materials.

How to apply and contacts

Apply online via the F6S platform at the specified competition page:register on F6S, complete the application form and upload supporting documents prior to the deadline. For technical or platform support contact: support@f6s.com. For challenge-specific queries consult the AI-BOOST website and challenge annex documents linked from the AI-BOOST site. Evaluator application deadline for external experts is 29 July 2026, 17:00 CEST, with remuneration for application review indicated in the call documents.

Relevant links and references:Official call listing on EC portal and AI-BOOST challenge annex PDFs are the authoritative sources for requirements and detailed technical annexes. See AI-BOOST website and F6S competition page for application forms and guideline documents. Additional reference: European Commission Funding & Tenders Portal listing for the competitive call AI-BOOST call on EC Portal AI-BOOST website F6S competition page Challenge 1 PDF Challenge 2 PDF ¹

Comprehensive summary

What is this opportunity about and how to explain it:The AI Challenge Competition under the AI-BOOST initiative is an EU-funded, open innovation competition designed to accelerate Generative AI solutions across strategic industrial and societal domains. It launches four targeted challenges, each led by an industry or research owner that provides datasets, mentoring and evaluation frameworks. The competition uses a two-stage funnel: a Spark Phase where concept notes compete for five prize slots per challenge (€28,500 each), followed by an Advance Phase where selected teams develop and validate prototypes over five months and compete in a live final for a €100,000 grand prize per challenge. The programme combines monetary prizes with significant technical and infrastructural support, including optional HPC access, real-world testbeds, and co-creation with Challenge Owners. Eligible applicants include individuals, SMEs, startups, universities, research institutes and consortia established in EU Member States or eligible Associated Countries. Applicants must submit via the F6S platform and follow the detailed guidelines provided by AI-BOOST. The competition emphasises responsible AI, reproducible evaluation against challenge-specific KPIs, safety and human oversight, and aims to advance TRL levels to facilitate real-world adoption of successful solutions.

Footnotes

1. 1F6S is the mandatory submission portal for proposals. The F6S competition page contains the formal application form and upload requirements. Contact support@f6s.com for platform assistance.

Competition Structure and Prizes

Challenge 1: GenAI for Autonomous Robots in Agriculture

This challenge, led by Consorzio Intellimech and JOiNT LAB, aims to democratize robotic programming by translating natural language into executable ROS-compatible commands using Vision-Language-Action (VLA) models ¹.

Target Outcome:Participants will develop a modular Proof of Concept at Technology Readiness Level (TRL) 5, initially focused on enhancing vineyard productivity but applicable to industrial maintenance and field operations ¹.

1. 1Intent Recognition Accuracy: ≥90%
2. 2Task Planning Success Rate: ≥85%
3. 3Robustness to Language Variability: ≥85%
4. 4End-to-End Latency: ≤5–10 seconds (Optional)

Challenge 2: Agentic AI for Automated CAD Generation

Led by SIAD Group, this challenge focuses on developing an agentic AI system to automate the design and simulation of industrial piping, specifically routing compressor tubes within skids ².

Target Outcome:The solution must generate parametric 3D CAD models from natural language and 2D drawings, automating mesh generation and convergence analysis to reach TRL 5–6 ².

1. 1Engineering Effort Reduction: AI hours < Human hours (Ratio < 1)
2. 2Package Size Reduction: AI-generated dimensions < Baseline dimensions (Ratio < 1)

Eligibility and Application Process

Eligible applicants include individuals (students, researchers), SMEs (including startups), academia, universities, NGOs, and consortia established in EU Member States or Associated Countries ³.

Key Requirements:Participants must have technical capacity and experience working on AI systems, and projects must focus exclusively on civil applications ³.

1. 1Register on the F6S platform via the AI-BOOST website
2. 2Complete the full application form (all mandatory fields)
3. 3Submit the Concept Note before 25 August 2026, 17:00 CEST
4. 4Review Guidelines for Applicants and full Challenge Descriptions

Timeline and Support

The competition follows a two-phase funnel:the SPARK Phase (Concept Note submission) and the ADVANCE Phase (five-month development and validation) ³.

Participants receive technical guidance, access to challenge-specific resources, optional access to CINECA’s LEONARDO supercomputer, and opportunities to collaborate with Challenge Owners ³.

The Final Live Event is scheduled for February 2027 in Brussels, where winners will pitch and demonstrate their solutions ³.

Footnotes

1. 1Challenge 1 Description: AI-BOOST Challenge 1 PDF
2. 2Challenge 2 Description: AI-BOOST Challenge 2 PDF
3. 3Official Competition Page: AI Challenge Competition and Application Portal: F6S Apply