From Literature Search to Submission: An End-to-End Agent Research Workflow with SciSpace
In this session, Marija Đukić (University of Belgrade, Faculty of Organizational Sciences) walks through the exact workflow behind her recently published research on Big Data Analytics maturity models showing, step by step, how SciSpace was used as a cognitive amplifier across the most demanding stages of academic work.
Her paper, presented at FedCSIS 2026, involved synthesising evidence across dozens of systematic literature reviews, generating and critically evaluating a 121-criterion assessment hierarchy, and iterating that hierarchy through multiple human–AI refinement cycles. She'll show you how SciSpace handled each of these tasks and, critically, where human judgment had to take over.
You'll walk away knowing how to:
Use SciSpace to synthesise evidence from large bodies of literature into a structured, usable framework
Run iterative AI-assisted evaluation cycles on your own research artifacts
Prompt effectively so AI justifies recommendations with literature — reducing hallucinations and making verification faster
Identify exactly where AI adds value (knowledge synthesis, conceptual consistency) vs. where it falls short (practical relevance, methodological judgment)
Who this is for: PhD researchers, academics, and research professionals working on literature-heavy, knowledge-intensive projects — systematic reviews, framework development, model validation, or any work where you're synthesising large bodies of evidence into original contributions.
Based on: "Human–AI Collaboration in Maturity Model Development: Evidence from Big Data Analytics Assessment Criteria" — Đukić, Dujmović & Luković, FedCSIS 2026.