Research that informs engineering.
AFC's R&D activity explores how intelligent systems can learn, adapt and remain dependable as their environment changes.
The objective is practical: understand the mechanisms that make long-lived AI systems safer, more measurable and easier to control.
Research areas
R&D / 01–06Systems capable of adapting to changing data, environments and requirements while maintaining controlled behaviour.
Incremental learning strategies for systems that must keep improving after deployment.
Methods for learning new information without unnecessarily degrading previously acquired capabilities.
Detecting drift, regression and silent degradation in systems that operate over long periods.
Architectural patterns that keep AI components observable, replaceable and testable.
Permission models and data boundaries for systems that reason over sensitive information.
Why a small engineering company does research at all.
Most production AI problems are not model-selection problems. They appear later: behaviour drifts as data changes, new capabilities degrade old ones, quality becomes hard to measure, and control weakens exactly when the system becomes useful.
Those are research questions with immediate engineering consequences. Working on them deliberately — rather than discovering them during an incident — is what keeps AFC's architectural decisions defensible.
Independent research by AFC's founder
Ivan Zdravkov's doctoral research at the Institute of Robotics, Bulgarian Academy of Sciences, focuses on adaptive large language models for social robotics through continual learning and knowledge retention mechanisms.
The academic research remains independent from AFC's commercial work, while the underlying questions — adaptation, stability and reliable long-term behaviour — directly inform AFC's engineering perspective.
Explore the research at zdravkov.info ↗The doctoral research is conducted independently. It does not imply endorsement, partnership, funding or commercial participation by the Bulgarian Academy of Sciences or the Institute of Robotics.