Qwestor is an autonomous research assistant and agentic research platform being developed to support end-to-end research workflows. Rather than functioning only as a question-answering system, Qwestor works with users to understand research problems, develop and execute plans, retrieve and organize evidence, evaluate results, revise approaches when necessary, and produce structured reports.
Its architecture combines practical product capabilities, including conversational research, persistent context, planning, retrieval, tool use, document processing, and multi-agent orchestration, with neuro-symbolic cognitive mechanisms. These include PLN-RAG for symbolic and contradiction-sensitive reasoning, motivation-based action selection, memory, and inspectable decision processes.
Qwestor is being developed toward a usable MVP while also serving as a controlled environment for testing AGI-relevant cognitive mechanisms in realistic workflows.