LECISONS brings together cognitive AI research, software engineering and practical implementation. Our two research programmes address complementary challenges: auditable decision-making and adaptive reasoning.
Auditable cognitive architecture and industrial validation.
Bidirectional abstraction and reasoning under uncertainty.
Research and project development · proposed outcomes remain subject to validation.
We investigate how AI can make decisions that can be examined, and how it can revise its understanding when the evidence changes. EVE-ARC and DeepARC pursue these questions through separate research scopes and evaluation plans.
A neuro-symbolic architecture combining learned world models, structured memory, planning and objective verification, with a focus on industrial integration and traceable decisions.
A research mechanism through which abstractions and structured reasoning would improve one another, with controlled concept revision when evidence is incomplete, conflicting or changing.
The programmes have distinct research questions, activities and intended results. Background architecture work is distinguished from new project-generated research.
Auditable cognitive AI for complex environments
EVE-ARC is a project proposal in the IPCEI AI context. It investigates how a cognitive architecture can connect learned models of an environment with structured memory, planning and explicit checks on proposed objectives.
Investigate learned world models, structured memory and planning as parts of an integrated neuro-symbolic system.
Develop objective-verification methods and replayable decision records within a defined system boundary, so decisions can be examined and challenged.
Study how to connect the architecture with industrial data and enterprise systems while preserving traceability, security controls and human oversight.
Plan validation in industrial reference environments and a bounded human-centred decision-support setting. Build evidence of performance, limitations and deployment readiness.
An integrated research system, reusable verification and integration methods, and documented validation evidence. These are project objectives; commercial readiness and certification are not presented as achieved results.
IPCEI AI is the project context. This presentation does not imply an approved funding award or formal portfolio status.
Bidirectional Abstraction and Reasoning for Trustworthy Cognitive AI
DeepARC is a proposed research programme investigating a two-way relationship between learned abstractions and structured reasoning. Reasoning outcomes would trigger controlled revisions of concepts; revised concepts would then guide further reasoning.
Can this two-way loop improve adaptation and transfer when evidence is incomplete, conflicting or changing?
Investigate typed feedback, uncertainty-aware revision and mechanical checks within a restricted, bounded language, with a traceable record of each change.
Compare the mechanism with fixed-symbolic, learned-only and one-way baselines, using controlled tasks and a laboratory demonstrator.
A testable mechanism, reproducible evaluation methods and evidence of both benefits and limitations. The application domain will be selected with partners.
DeepARC is new research, drawing on background cognitive-AI architecture work (CAIRN/EVE). The proposed capabilities are not yet demonstrated. TRL 4 is a research target, not a statement of current maturity.
We welcome conversations with research groups, technology organisations and application owners. Tell us which programme you are interested in and what expertise, environment or evaluation capability you can contribute.
Discuss industrial validation, enterprise integration, verification methods and traceable decision-support environments. Enquiries will be considered in relation to the existing project scope and consortium arrangements.
We seek scientific co-design partners in neuro-symbolic AI, reasoning under uncertainty, formal methods and independent evaluation, plus an application or validation partner. The consortium is being formed; external partners are not yet confirmed.
Research summaries are available on request. Initial discussions can focus on scientific fit, responsibilities and validation resources.
Alongside research, LECISONS works on AI implementation, automation and digital growth. Commercial engagements are scoped separately from research proposals.
Connect customer interactions, CRM, order flows and back-office processes.
Connect AI capabilities with existing software, APIs and operational workflows.
Positioning, acquisition and conversion support help digital products reach customers. These services complement the engineering work.
Based in Ljubljana, Slovenia, LECISONS d.o.o. combines cognitive-AI research with software engineering, systems integration and commercial implementation. We connect long-term research questions with the practical work required to build and evaluate useful systems.
Our research approach prioritises explicit hypotheses, reproducible evaluation and traceability. Background intellectual property and new project results are treated separately.
CEO · business enquiries and consortium collaboration
LinkedInFor research or technology collaboration, mention EVE-ARC or DeepARC and your potential contribution. For an implementation project, describe the business process you want to improve.
Email LECISONSA scientific summary is available on request.