Horizon Europe / Research proposal
VERITAS-CA
Shared decisions. Understandable reasons.
How can AI agents work together when they see different parts of a situation and cannot always communicate? VERITAS-CA proposes to investigate that question through trustworthy reasoning and planning.
Submitted on 28 October 2026 to the EIC Pathfinder Challenges DeepRAP call. Funding and evaluation outcomes have not been announced here.
Explore the proposal
Research question
Working together without a complete picture.
Imagine two inspection agents looking at different parts of a facility. One has a recent observation; the other has an older report. A delayed message can leave them with different understandings of the same task. The question is how to coordinate while keeping those differences and uncertainties visible.
VERITAS-CA stands for Verifiable Explainable Resilient Trustworthy Autonomous Cognitive Agents. It is a Neura Parse research proposal about the reasoning behind a shared decision, the information available to each agent and the evidence a reviewer would need to understand what happened.
The inspection example explains the research problem. It does not describe a customer deployment or a completed field trial.
Different views
Each agent has only part of the picture.
A shared task
Reasoning and planning must account for uncertainty.
A decision to examine
The basis and limits should remain understandable.
Proposed approach
Reasoning, planning and evidence, studied together.
The proposal brings several research questions into one evaluation. These are intended directions, with their value and limits still to be tested.
- Reason with incomplete information
- An agent sees part of a situation, rather than an unquestionable picture of the world. The research asks how uncertainty can remain visible when a decision is made.
- Plan with limited communication
- Different observations and delayed messages can complicate a shared task. VERITAS-CA proposes to investigate coordination under these conditions in controlled research settings.
- Make decisions reviewable
- An explanation should help a reviewer understand the basis of a decision. The proposed work connects decision-making with evidence that can be examined and revisited.
- Measure the trade-offs
- A useful method must be judged alongside the time, communication and computing resources it needs. Strong reference methods and unsuccessful cases belong in that comparison.
Submission record
A proposal submitted. A research outcome still ahead.
The submission milestone is confirmed by the European Commission’s Funding & Tenders Portal notification. It is separate from any later evaluation, funding or implementation decision.
- Proposal
- VERITAS-CA
- Applicant
- Neura Parse Ltd, United Kingdom
- Programme
- Horizon Europe / EIC Pathfinder Challenges 2026
- Challenge
- DeepRAP: Deep Reasoning, Abstraction & Planning towards trustworthy Cognitive AI Systems
- Submitted
- 28 October 2026
- Status
- Submitted proposal; no funding award announced
The programme link describes the call. It is not an award record for VERITAS-CA.
Evaluation and limits
Evidence before stronger claims.
The proposed evaluation concerns controlled laboratory software and simulated civil inspection or maintenance tasks. It would examine decision quality, communication demands and computing cost under defined conditions. A laboratory validation is a research objective, not a result already achieved by this proposal.
Classical and quantum-assisted inference would be compared within a bounded research question. A useful negative result, including cases where a simpler classical method is preferable, is part of the intended scientific value.
This scope does not establish certified safety, unrestricted autonomy, a live infrastructure pilot or a general performance advantage. Consequential actions remain subject to human authority and the safety requirements of any future application.
Related public QANTIS research provides background on quantum-assisted belief updates. Those published studies are separate from VERITAS-CA and do not demonstrate that its proposed multi-agent system has been completed.
Read the separate QANTIS research recordQuestions and further reading
What this milestone means.
Has VERITAS-CA received Horizon Europe funding?
The proposal was submitted on 28 October 2026. Submission does not establish selection for evaluation, an award, a grant agreement or European Commission endorsement. No funding outcome is announced on this page.
Is this an available product or a deployed system?
VERITAS-CA is a research proposal. It describes work to be investigated and evaluated, rather than a released product, a completed laboratory result or an operational deployment.
Where does quantum research fit?
The proposal includes a bounded comparison of classical and quantum-assisted inference. Quantum advantage is a research question, not a promised outcome. Planning and consequential authority remain separate from that comparison.
What could be shared as the research develops?
Appropriate research findings and reproducibility materials may be shared subject to rights, confidentiality and any applicable funding conditions. This page makes no claim that a VERITAS-CA dataset, code release or experimental result is already available.
Public overview updated on 29 October 2026. Proposal documents, personal information and protected technical or financial details are not part of this publication.