Loading…
Strategic Analysis
This EIC Pathfinder Challenge funds novel approaches that go beyond current, traditional deep learning and reinforcement learning paradigms to significantly improve the Reasoning, Abstraction and Planning capabilities of AI systems. Each proposal addresses one or more of the three capabilities (deep reasoning, deep abstraction, deep planning) and delivers models or architectures that handle multimodal data and uncertainty under constrained computational resources, provable trustworthiness mechanisms aligned with ethical and legal standards including the EU AI Act, and a demonstration integrated in a cognitive AI system reaching TRL 4 on complex real-world tasks. Proposals also propose new methods and metrics for evaluating and certifying reasoning and trustworthiness, follow the FAIR principles and develop synergies with EU initiatives such as TEFs, eBrains, RAISE, the AI-on-demand Platform and the Quantum Flagship. The portfolio is built to cover all three capabilities, a variety of technological approaches and application domains, with a preference for explainability and formal guarantees.
TRL 2 → 4
Grouped by cognitive capabilities addressed (portfolio category 1) — a proposal targets one branch.
Deep Reasoning: Moving beyond statistical pattern matching to support causal inference, logical reasoning, and context-aware or commonsense decision-making in complex, unstructured environments. Deep reasoning systems should be able to reconcile multiple sources of information, provide transparent and explainable rationales for their outputs, and align with human values and expectations, ensuring trustworthy and accountable operations in demanding real-world scenarios
Deep Abstraction: Enabling AI systems to generalise insights from limited data by forming, manipulating, and refining high-level concepts, analogies, and representations that can be transferred across diverse application domains. This includes the development of internal world models to support abstraction, foster commonsense understanding, and integrate semantic and contextual awareness
Deep Planning: Developing robust, adaptive, and scalable planning algorithms/models capable of operating in open-world, agentic, or uncertain real-time environments. Emphasis is placed on long-term, flexible planning; approaches should explore hierarchical planning across multiple temporal levels, contingency planning for effective fallback strategies, and continual re-planning to dynamically update plans as environments evolve
Innovative ideas put forward under this Challenge must explore novel approaches, including combinations of existing techniques (i.e. neuro-symbolic AI), or the creation of entirely new frameworks that go beyond current, traditional, deep learning and reinforcement learning paradigms. These could be inspired by developments in diverse fields such as neuroscience, biology, physics, philosophy and more
Models and/or architectures that handle multimodal data and knowledge, uncertainty, and can be trained and deployed with constrained computational resources
Provable trustworthiness mechanisms ensuring explainability, transparency, fairness, risk evaluation, security and alignment with ethical and legal standards, including fundamental rights and the EU AI Act
Demonstrate the developed capabilities integrated in a cognitive AI system (reaching TRL4) performing complex real-world tasks (e.g., scientific discovery, decision support, problem solving) as well as simulations at a scale
Propose new methods and metrics for evaluating and certifying reasoning and trustworthiness in AI as well as the use of the computational resources
Advance the scientific state-of-the-art and build a robust, interoperable, and application-driven community, positioning Europe at the forefront of trustworthy cognitive AI
Lay the foundations for future European leadership in safe, human-centric cognitive AI, supporting sovereignty and competitiveness in key sectors
Support the ambitions of the AI Act and the European approach to Artificial Intelligence
AI Act
highThe AI Act establishes a regulatory framework for AI systems, categorising them by risk and setting requirements for transparency, safety, and accountability, particularly for high-risk applications like healthcare.
The second expected outcome asks for provable trustworthiness mechanisms ensuring explainability, transparency, fairness, risk evaluation, security and alignment with ethical and legal standards, including fundamental rights and the EU AI Act. Evaluators will look for how the proposed mechanisms achieve this.
European approach to Artificial Intelligence
highThe European approach to Artificial Intelligence aims to foster AI excellence and trust by promoting human-centric, ethical, and secure AI technologies. It sets strategic priorities to strengthen EU research and industrial capacity while ensuring fundamental rights and safety standards are safeguarded.
Evaluators expect proposals to embed European values and ethics by design, demonstrating how the cognitive AI systems developed are trustworthy, transparent, explainable, and aligned with the EU Coordinated Plan on AI.
Specific requirements
Single applicant
A single legal entity may apply if it is established in a Member State or an Associated Country, unless the Challenge states otherwise. In single beneficiary projects, mid-caps and larger companies are not permitted.
Consortium of two entities
Two independent legal entities from two different Member States or Associated Countries.
Consortium of three or more entities
At least three independent legal entities, each established in a different country: at least one in a Member State, and at least two others each in different Member States or Associated Countries.
Entities eligible for funding
Entities established in a Member State (including outermost regions and linked Overseas Countries and Territories), in a country associated to Horizon Europe Pillar III, or in one of the low- and middle-income countries listed in Annex 2, section B.2 of the EIC Work Programme 2026. Other entities may participate without funding, unless their participation is considered essential.
Exclusive focus on civil applications
Projects must focus exclusively on civil applications, and must not aim at human cloning for reproductive purposes, heritable modification of the genetic heritage of human beings, or the creation of human embryos solely for research or stem cell procurement (Annex 2, B.4).
Gender equality plan
Public bodies, research organisations and higher education establishments from Member States and Associated Countries must have a gender equality plan; a self-declaration is requested at proposal stage. This does not apply to private for-profit organisations, including SMEs (Annex 2, B.8).
Protection of European communication networks
Applications with elements that concern the evolution of European communication networks (5G, post-5G and related technologies) are subject to restrictions: entities assessed as high-risk suppliers of mobile network communication equipment are not eligible (Annex 2, B.1).
EU contribution
An EU contribution of up to EUR 4 million is considered appropriate; larger amounts may be requested if duly justified. Funding rate 100%, as a lump sum.
In order to apply, your proposal must meet the general eligibility requirements (see Annex 2 of EIC Work Programme 2026) as well as specific eligibility requirements for the Challenge (please see the Topic description above).
Please check for particular elements (e.g., specific application focus or technology) in the respective Challenge chapter.
The EIC Pathfinder Challenges support collaborative or individual research and innovation from consortia or from single legal entities established in a Member State or an Associated Country (unless stated otherwise in the specific Challenge chapter). In case of a consortium your proposal must be submitted by the coordinator on behalf of the consortium. Consortia of two entities must be comprised of independent legal entities from two different Member States or Associated Countries. Consortia of three or more entities must include as beneficiaries at least three legal entities, independent from each other and each established in a different country as follows:
The legal entities may for example be universities, research organisations, SMEs, start-ups, natural persons. In the case of single beneficiary projects, mid-caps and larger companies will not be permitted.
Applications with elements that concern the evolution of European communication networks (5G, post-5G and other technologies linked to the evolution of European communication networks) will be subject to restriction for the protection of European communication networks (see Annex II – Section B1).
The standard admissibility and eligibility conditions and the eligibility of applicants from third countries are detailed in Annex 2.
Proposal page limit and layout:
Described in Part B of the Application Form available in the Submission System.
Sections 1 to 3 of the part B of your proposal, corresponding respectively to the evaluation criteria Excellence, Impact, and Quality and Efficiency of the Implementation, must consist of a maximum of 30 format A4 pages. Excess pages will be automatically made invisible, and will not be taken into consideration by the evaluators. Please also consult Annex 2 of the EIC Work Programme 2026.
Described in Annex 2 of the EIC Work Programme 2026.
Described in Annex 2 of the EIC Work Programme 2026.
Described in Annex 2 of the EIC Work Programme 2026.
Described in Section II of the EIC Work Programme 2026.
Described in Section II of the EIC Work Programme 2026 and the Online Manual.
Described in Section II of the EIC Work Programme 2026.
Please refer to the Lump Sum Model Grant Agreement (Lump Sum MGA) used for Lump Sum EIC actions under Horizon Europe.
Described in the EIC Work Programme 2026.
Frequently Asked Questions (FAQs)
Standard application form (HE EIC Pathfinder Challenges)
Standard evaluation form (HE EIC Pathfinder Challenges)
Challenge Guide: Deep Reasoning, Abstraction & Planning towards trustworthy Cognitive AI Systems
Call-specific instructions
Information on clinical studies (HE)
Guidance: "Lump sums - what do I need to know?"
HE Framework Programme 2021/695
HE Specific Programme Decision 2021/764
EU Financial Regulation 2018/1046
Rules for Legal Entity Validation, LEAR Appointment and Financial Capacity Assessment
EU Grants AGA — Annotated Model Grant Agreement
Funding & Tenders Portal Online Manual
Source: EIC Work Programme 2026, p. 34-38; Challenge Guide, section 3.
Each criterion is scored from 0 to 5. The Work Programme sets a threshold per criterion and no overall threshold; the weights serve the ranking.
| Criterion | Threshold | Weight | Aspects assessed |
|---|---|---|---|
| Excellence | 4/5 | 50% | Objectives and relevance to the Challenge: How clear are the project's objectives? How relevant are they in contributing to the overall goal and the specific objectives of the Challenge? Novelty: To what extent is the proposed work ambitious and goes beyond the state-of-the-art? Plausibility of the methodology: How sound is the proposed methodology, including the underlying concepts, models, assumptions, appropriate consideration of the gender dimension in research content, and the quality of open science practices? |
| Impact | 3.5/5 | 30% | Potential Impact: How credible are the pathways to achieve the expected outcomes and impacts of the Challenge? To what extent would the successful completion of the project contribute to this? Innovation potential: How realistic is the proof of principle for demonstrating the potential impact of the technology for the challenge? How adequate are the proposed measures for protection of results and any other exploitation measures to facilitate future translation of research results into innovations with positive societal, economic or environmental impact? How suitable are the proposed measures for involving and empowering key actors that have the potential to take the lead in translating research into innovations in the future? Communication and Dissemination: How suitable are the proposed measures, including communication activities, to maximise expected outcomes and impacts for raising awareness about the project results' potential to establish new markets and/or address global challenges? |
| Quality and efficiency of the implementation | 3/5 | 20% | Work plan: How coherent and effective are the work plan (work packages, tasks, deliverables, milestones, timeline, etc.) and risk mitigation measures in order to achieve the project objectives? Allocation of resources: How appropriate and effective is the allocation of resources (comprising person-months and other cost items) to work packages and consortium members? Quality of the applicant/consortium (depends if mono or multi-beneficiaries): To what extent does the applicant / do all consortium members have the necessary capacity and high quality expertise for performing the project tasks? |
All proposals that meet the thresholds are considered together. The evaluation committee maps them on the categories of the Challenge Guide and selects a portfolio that achieves the expected outcomes and impacts of the Challenge. A proposal with a higher score may not be retained, and the committee may propose minor adjustments for the consistency of the portfolio. The portfolio paragraph of section 1.1 and the self-assessment table feed this step.
Source: EIC Work Programme 2026, p. 35-37 and Annex 2 (A.2, F.1); Part B form v4.3; Evaluation form for EIC Pathfinder Challenges.
Source: EIC Work Programme 2026, p. 33-34 and Annex 2 (E.1); Part B form v4.3, tables for section 3.1; Challenge Guide, section 5.
The selected projects will also be assigned to lead and/or engage in portfolio activities centred on the following priorities:
A steering committee steered by the Programme Manager and four working groups, each consortium nominating a representative per group: WG1 Technology integration, validation and demonstration (DeepRAP benchmark, interoperability standards, cross-project validation, hybrid systems, joint pilots); WG2 Transition of technology to innovation; WG3 Regulation, ethics and trustworthiness; WG4 Communication and dissemination. Online meetings about every three months and one annual meeting in person.
Source: EIC Work Programme 2026, p. 47; Challenge Guide, section 4.3.
The Work Programme advises a work package dedicated to portfolio activities with at least 10 person-months. The Challenge Guide provides a template (Annex 1): WPX Portfolio management, from month 1 to the end of the project, with a task on portfolio governance and working groups, a task on the portfolio strategic plan and common documents, and one deliverable per reporting period, Deliverable X.1.i: Report on portfolio activities (type R, dissemination level SEN). You may describe concrete activities or remain generic. Without this work package the proposal is not scored lower, but it will be requested at grant preparation, within the same maximum grant.
| Project duration (months) | Number of periods | Period durations (months) |
|---|---|---|
| 12 | 1 | 12 |
| 18 | 1 | 18 |
| 24 | 2 | 12, 12 |
| 30 | 2 | 12, 18 |
| 36 | 2 | 12, 24 |
| 42 | 3 | 12, 12, 18 |
| 48 | 3 | 12, 18, 18 |
| 60 | 4 | 12, 16, 16, 16 |
Each work package should be a substantial part of the work plan; payments follow the work packages completed within each reporting period.
Source: EIC Work Programme 2026, p. 34; Challenge Guide, section 4.1 and Annex 1; Part B form v4.3, section 3.1.
Source: EIC Work Programme 2026, Annex 6, section 2.
Everything the call asks for, seen from the call's point of view. Each line shows what answers it, and which partner carries it.
This matrix lists everything the call asks for: outcomes, impacts, scope, the requirements buried in the call text, and policy alignment. Sign up free and GrantForge tracks each line against the concept you build.
| Requirement | Covered by | Carried | Status |
|---|---|---|---|
| Specific objectives | |||
| SC1Innovative ideas put forward under this Challenge must explore novel approaches, including combinations of existing techniques (i.e. neuro-symbolic AI), or the creation of entirely new frameworks that go beyond current, traditional, deep learning and reinforcement learning paradigms. These could be inspired by developments in diverse fields such as neuroscience, biology, physics, philosophy and more | · | · | Sign up to track |
| SC2Deep Reasoning: Moving beyond statistical pattern matching to support causal inference, logical reasoning, and context-aware or commonsense decision-making in complex, unstructured environments. Deep reasoning systems should be able to reconcile multiple sources of information, provide transparent and explainable rationales for their outputs, and align with human values and expectations, ensuring trustworthy and accountable operations in demanding real-world scenarios | · | · | Sign up to track |
| SC3Deep Abstraction: Enabling AI systems to generalise insights from limited data by forming, manipulating, and refining high-level concepts, analogies, and representations that can be transferred across diverse application domains. This includes the development of internal world models to support abstraction, foster commonsense understanding, and integrate semantic and contextual awareness | · | · | Sign up to track |
| SC4Deep Planning: Developing robust, adaptive, and scalable planning algorithms/models capable of operating in open-world, agentic, or uncertain real-time environments. Emphasis is placed on long-term, flexible planning; approaches should explore hierarchical planning across multiple temporal levels, contingency planning for effective fallback strategies, and continual re-planning to dynamically update plans as environments evolve | · | · | Sign up to track |
| Expected outcomes | |||
| EO1Models and/or architectures that handle multimodal data and knowledge, uncertainty, and can be trained and deployed with constrained computational resources | · | · | Sign up to track |
| EO2Provable trustworthiness mechanisms ensuring explainability, transparency, fairness, risk evaluation, security and alignment with ethical and legal standards, including fundamental rights and the EU AI Act | · | · | Sign up to track |
| EO3Demonstrate the developed capabilities integrated in a cognitive AI system (reaching TRL4) performing complex real-world tasks (e.g., scientific discovery, decision support, problem solving) as well as simulations at a scale | · | · | Sign up to track |
| EO4Propose new methods and metrics for evaluating and certifying reasoning and trustworthiness in AI as well as the use of the computational resources | · | · | Sign up to track |
| Other requirements | |||
| REQ1Data, models and results following the FAIR principlesFollow the FAIR principles ensuring all data, models, and results are Findable, Accessible, Interoperable, and Reusable to maximise transparency, reproducibility, and impact. | · | · | Sign up to track |
| REQ2Synergies with EU AI initiatives (TEFs, eBrains, RAISE, AIoD, Quantum Flagship)Develop synergies with EU initiatives such as TEFs (AI Testing and Experimentation Facilities), eBrains, Resource for AI Science in Europe (RAISE), AI-on-demand Platform (AIoD) and the Quantum Flagship. | · | · | Sign up to track |
| REQ3Work package dedicated to portfolio activities (at least 10 person-months advised)The Work Programme advises a work package dedicated to portfolio activities with at least 10 person-months, and the Challenge Guide provides a template for it. Without it the proposal is not scored lower, but the work package is requested at grant preparation within the same maximum grant. | · | · | Sign up to track |
| Expected impacts | |||
| EI1Advance the scientific state-of-the-art and build a robust, interoperable, and application-driven community, positioning Europe at the forefront of trustworthy cognitive AI | · | · | Sign up to track |
| EI2Lay the foundations for future European leadership in safe, human-centric cognitive AI, supporting sovereignty and competitiveness in key sectors | · | · | Sign up to track |
| EI3Support the ambitions of the AI Act and the European approach to Artificial Intelligence | · | · | Sign up to track |
| Underlying policies | |||
| POL1ai actThe AI Act establishes a regulatory framework for AI systems, categorising them by risk and setting requirements for transparency, safety, and accountability, particularly for high-risk applications like healthcare. | · | · | Sign up to track |
| POL2european approach to artificial intelligenceThe European approach to Artificial Intelligence aims to foster AI excellence and trust by promoting human-centric, ethical, and secure AI technologies. It sets strategic priorities to strengthen EU research and industrial capacity while ensuring fundamental rights and safety standards are safeguarded. | · | · | Sign up to track |
The binding rules of this call. Items marked auto are verified by GrantForge from the call and the template. The others are yours to confirm.
LMIC entities auto-eligible
Low/middle-income country entities are automatically eligible for funding.
Single applicant
A single legal entity may apply if it is established in a Member State or an Associated Country, unless the Challenge states otherwise. In single beneficiary projects, mid-caps and larger companies are not permitted.
Consortium of two entities
Two independent legal entities from two different Member States or Associated Countries.
Consortium of three or more entities
At least three independent legal entities, each established in a different country: at least one in a Member State, and at least two others each in different Member States or Associated Countries.
Entities eligible for funding
Entities established in a Member State (including outermost regions and linked Overseas Countries and Territories), in a country associated to Horizon Europe Pillar III, or in one of the low- and middle-income countries listed in Annex 2, section B.2 of the EIC Work Programme 2026. Other entities may participate without funding, unless their participation is considered essential.
Exclusive focus on civil applications
Projects must focus exclusively on civil applications, and must not aim at human cloning for reproductive purposes, heritable modification of the genetic heritage of human beings, or the creation of human embryos solely for research or stem cell procurement (Annex 2, B.4).
Gender equality plan
Public bodies, research organisations and higher education establishments from Member States and Associated Countries must have a gender equality plan; a self-declaration is requested at proposal stage. This does not apply to private for-profit organisations, including SMEs (Annex 2, B.8).
Protection of European communication networks
Applications with elements that concern the evolution of European communication networks (5G, post-5G and related technologies) are subject to restrictions: entities assessed as high-risk suppliers of mobile network communication equipment are not eligible (Annex 2, B.1).
EU contribution
An EU contribution of up to EUR 4 million is considered appropriate; larger amounts may be requested if duly justified. Funding rate 100%, as a lump sum.
The call requires you to pick or distribute several structuring elements before submission. These are locked at proposal creation time and drive the project backbone. Read-only here.
deeprap-capabilities
Address one or more of the three cognitive capabilities. Choosing two or three places the project in the Multi-Capability category (MC) of the Challenge Guide, which requires clear integration mechanisms and which the portfolio building prioritises.
Deep Reasoning (DR)
Causal inference, logical and probabilistic reasoning, context-aware decision-making, commonsense reasoning, end-to-end differentiable reasoning, dynamic, parallel and multi-modal reasoning (including visual, spatial, temporal), explainable AI frameworks.
Deep Abstraction (DA)
Generalisation from limited data, concept formation, analogical reasoning, abstract world model building, semantic understanding and context recognition, transfer learning.
Deep Planning (DP)
Adaptive planning algorithms, real-time decision making, hierarchical long-term planning at multiple levels of temporal abstraction, contingency planning, continual re-planning.
deeprap-approach
One project may address several technological approaches. The portfolio looks for diversity while prioritising hybrid approaches; for trustworthiness, explainability and formal guarantees are preferred.
Neuro-symbolic AI (NeSy)
Hybrid approaches combining neural networks with symbolic reasoning.
Advanced Deep Learning (ADL)
Novel neural architectures, transformer variants, attention mechanisms or other deep learning innovations.
Reinforcement Learning (RL)
Advanced RL approaches, multi-agent RL, or hierarchical RL.
Cognitive Architectures (CA)
BDI-based systems, cognitive modelling approaches, or biologically inspired architectures.
Novel Interdisciplinary Frameworks (NIF)
Approaches inspired by neuroscience, biology, physics, philosophy, or other fields.
Formal Methods Integration (FMI)
Approaches incorporating formal verification, logic programming, or mathematical guarantees.
Multimodal Integration (MMI)
Systems handling multiple data types and modalities (text, vision, audio, sensors).
Trustworthiness mechanisms Integration (TM)
Explainability and interpretability, fairness, safety verification, real world robustness.
Other approach
deeprap-domain
Projects are mapped by their primary application focus. Pick the primary domain first, and optionally a secondary focus, as in the self-assessment table of the Challenge Guide.
Scientific Discovery (SD)
AI systems for hypothesis generation, experiment design, or knowledge discovery (chemistry, materials, biology, climate etc.).
Decision Support Systems (DSS)
Applications in healthcare, finance, policy-making, or strategic planning.
Autonomous Systems (AS)
Robotics, autonomous vehicles, or other embodied AI applications.
Human-AI Collaboration (HAIC)
Systems designed for human-AI teaming or augmented intelligence.
Cybersecurity (CS)
AI systems for threat detection, response planning, or security analysis.
Industrial Applications (Ind)
Manufacturing optimization, supply chain management, or process control.
Social and Societal Applications (SSA)
AI systems addressing social challenges or public services.
deeprap-synergies
The Challenge Guide also categorises proposals on their capacity to contribute to portfolio-wide collaboration. Optional: tick the dimensions your project plans to contribute to.
Benchmark Development (Ben)
Contributing to portfolio-wide shared benchmarks for reasoning, abstraction and planning, including data, shared tasks and open evaluation protocols.
Interoperability (Int)
Developing or adopting standards, protocols and APIs for technical integration across the portfolio.
Joint Pilots and Demonstrations (JPD)
Multi-partner pilots combining results, tools or models from different projects, especially through multiagent or modular system integration.
5 key insights you must internalise before writing. Each is grounded in the call text and tells you what evaluators will actually look for. Share these with your consortium before drafting.
Every proposal must explore novel approaches, including combinations of existing techniques such as neuro-symbolic AI, or entirely new frameworks that go beyond current, traditional, deep learning and reinforcement learning paradigms. Position the approach explicitly against these paradigms in section 1.2.
Source: EIC Work Programme 2026, p. 45
Proposals address one or more of deep reasoning, deep abstraction and deep planning. The portfolio must cover all three, and the Challenge Guide prioritises projects addressing multiple capabilities, provided they bring clear integration mechanisms.
Source: EIC Work Programme 2026, p. 45; Challenge Guide, sections 3.1 and 3.2
Besides models or architectures working under constrained computational resources, the Challenge expects provable trustworthiness mechanisms aligned with ethical and legal standards including the EU AI Act, a demonstration integrated in a cognitive AI system reaching TRL 4 on complex real-world tasks, new methods and metrics for evaluating and certifying reasoning and trustworthiness, FAIR data, models and results, and synergies with EU initiatives such as TEFs, eBrains, RAISE, AIoD and the Quantum Flagship.
Source: EIC Work Programme 2026, p. 46
Proposals above all three thresholds (Excellence 4, Impact 3.5, Implementation 3) enter a second step where the evaluation committee classifies each one by cognitive capability, technological approach, application domain and synergy aspects, looking for breadth, commonality and a preference for explainability and formal guarantees and selects a coherent portfolio. A highly ranked proposal that has no synergy or commonality with the others may not be selected, and a proposal very similar to one already retained is set aside. Fill in the self-assessment table of the Challenge Guide in section 1.1.
Source: EIC Work Programme 2026, p. 35-37; Challenge Guide, sections 3 and 4.1
A work package dedicated to portfolio activities with at least 10 person-months is advised, and the Challenge Guide gives a template for it (Annex 1). Leaving it out does not lower the score, but it will be requested at grant preparation and financed from the same maximum grant: planning it now avoids cutting other work later.
Source: EIC Work Programme 2026, p. 34; Challenge Guide, section 4.1
Talk to the Grant Coach to build your concept. There is no set order: start wherever your project starts. The sections below mirror what the conversation produces, and your coverage tracks the progress. You can refine everything once your project workspace is created.
The branch you're building for — it shapes what the call expects of you, and what your coverage is measured against.
The call requires you to pick or distribute several structuring elements before submission. These are locked at proposal creation time and drive the project backbone. Read-only here.
deeprap-capabilities
Address one or more of the three cognitive capabilities. Choosing two or three places the project in the Multi-Capability category (MC) of the Challenge Guide, which requires clear integration mechanisms and which the portfolio building prioritises.
Deep Reasoning (DR)
Causal inference, logical and probabilistic reasoning, context-aware decision-making, commonsense reasoning, end-to-end differentiable reasoning, dynamic, parallel and multi-modal reasoning (including visual, spatial, temporal), explainable AI frameworks.
Deep Abstraction (DA)
Generalisation from limited data, concept formation, analogical reasoning, abstract world model building, semantic understanding and context recognition, transfer learning.
Deep Planning (DP)
Adaptive planning algorithms, real-time decision making, hierarchical long-term planning at multiple levels of temporal abstraction, contingency planning, continual re-planning.
deeprap-approach
One project may address several technological approaches. The portfolio looks for diversity while prioritising hybrid approaches; for trustworthiness, explainability and formal guarantees are preferred.
Neuro-symbolic AI (NeSy)
Hybrid approaches combining neural networks with symbolic reasoning.
Advanced Deep Learning (ADL)
Novel neural architectures, transformer variants, attention mechanisms or other deep learning innovations.
Reinforcement Learning (RL)
Advanced RL approaches, multi-agent RL, or hierarchical RL.
Cognitive Architectures (CA)
BDI-based systems, cognitive modelling approaches, or biologically inspired architectures.
Novel Interdisciplinary Frameworks (NIF)
Approaches inspired by neuroscience, biology, physics, philosophy, or other fields.
Formal Methods Integration (FMI)
Approaches incorporating formal verification, logic programming, or mathematical guarantees.
Multimodal Integration (MMI)
Systems handling multiple data types and modalities (text, vision, audio, sensors).
Trustworthiness mechanisms Integration (TM)
Explainability and interpretability, fairness, safety verification, real world robustness.
Other approach
deeprap-domain
Projects are mapped by their primary application focus. Pick the primary domain first, and optionally a secondary focus, as in the self-assessment table of the Challenge Guide.
Scientific Discovery (SD)
AI systems for hypothesis generation, experiment design, or knowledge discovery (chemistry, materials, biology, climate etc.).
Decision Support Systems (DSS)
Applications in healthcare, finance, policy-making, or strategic planning.
Autonomous Systems (AS)
Robotics, autonomous vehicles, or other embodied AI applications.
Human-AI Collaboration (HAIC)
Systems designed for human-AI teaming or augmented intelligence.
Cybersecurity (CS)
AI systems for threat detection, response planning, or security analysis.
Industrial Applications (Ind)
Manufacturing optimization, supply chain management, or process control.
Social and Societal Applications (SSA)
AI systems addressing social challenges or public services.
deeprap-synergies
The Challenge Guide also categorises proposals on their capacity to contribute to portfolio-wide collaboration. Optional: tick the dimensions your project plans to contribute to.
Benchmark Development (Ben)
Contributing to portfolio-wide shared benchmarks for reasoning, abstraction and planning, including data, shared tasks and open evaluation protocols.
Interoperability (Int)
Developing or adopting standards, protocols and APIs for technical integration across the portfolio.
Joint Pilots and Demonstrations (JPD)
Multi-partner pilots combining results, tools or models from different projects, especially through multiagent or modular system integration.
The problems this call frames, and who they affect. Your concept and plan address them.
Academic researchers and institutes advancing neuro-symbolic architectures, causal inference, and open-source benchmarks.
Developers and tech enterprises seeking energy-efficient, robust cognitive AI systems for autonomous agents and decision-support tools.
Standardization committees and EU AI Act compliance auditors evaluating AI safety, transparency, and auditability metrics.
Domain specialists requiring reliable, explainable decision-making in complex environments without risk of hallucination.
The long-term impacts your project should drive, and the policies they serve.
Establishment of novel foundational AI frameworks bridging symbolic reasoning and connectionist learning, reinforcing EU digital sovereignty.
Reduction in verification costs and commercialization lead-time for trustworthy cognitive AI solutions across European vertical industries.
Deployment of compute-efficient, verified cognitive agents reducing energy burdens and securing operations against systemic AI failures.