Ova četiri prompta pokrivaju delove literature review u kojima AI najviše štedi vreme bez gubitka strogosti: dizajn strategije pretrage iz fokusiranih research pitanja, ekstrakcija strukturisanih nalaza iz seta radova, sinteza preko izvora po temama i izgradnja gap analysis sa prioritizovanim preporukama. Svaki prompt treba popuniti sopstvenim kontekstom projekta i materijalom, a zatim ga pokrenuti u Claude-u, ChatGPT-ju, Gemini-ju ili bilo kom LLM-u sa dovoljno dugačkim kontekstualnim prozorom. Promptovi pretpostavljaju ljudskog istraživača u petlji — model pravi masovni prolaz, istraživač verifikuje citate, čita najuticajnije izvore u celosti i poseduje finalne preporuke.
Prompt 1: Fokusirana pretraga literature i ekstrakcija
You are an experienced research librarian helping me run a focused literature review for a UX project.
Project context: [tip proizvoda, korisnički segment, poslovno pitanje]
Research questions: [navedite 2-4 konkretna pitanja na koja obzir mora da odgovori]
Time window: [npr. poslednjih 5 godina za UX, duže za foundational studies]
Source types: [akademski radovi, industrijski izveštaji, samo peer-reviewed itd.]
For each research question:
1. Suggest 5-8 search queries (mix of academic phrasing and practitioner phrasing) that would surface the most relevant published evidence. Briefly justify each query.
2. List 3-5 high-credibility databases or sources to search for this specific question (Google Scholar, ACM Digital Library, Nielsen Norman Group, Baymard Institute, etc.).
3. Propose inclusion criteria for screening: what kinds of studies should be included or excluded for this question, and why.
4. Flag any keywords that are likely to surface a high volume of low-relevance noise, with suggestions for narrowing.
Format the output as a structured table with one row per research question and one column per recommended query, source, and inclusion criterion.Prompt 2: Ekstrakcija strukturisanih nalaza iz seta radova
I will paste a set of paper abstracts (or full-text PDFs if available) below. For each paper, extract a structured row with the following fields:
1. Citation (author, year, title, venue)
2. Research question(s) addressed by this paper
3. Methods used (sample size, study design, qualitative/quantitative/mixed)
4. Key findings, in 2-3 sentences, with direct quotes where possible
5. Limitations or context that might affect applicability to my project
6. Relevance score for my project (High / Medium / Low) and one-sentence justification
7. Implication for my design or research decision (one sentence — "if this finding holds, then...")
My project context: [tip proizvoda, korisnički segment, odluka koju će obzir informisati]
Papers to extract:
[ubacite abstrakte ili PDF]
After the table, list:
- Any paper where the abstract did not give enough information to extract reliably (flag for full-text reading)
- Any paper that seems methodologically weak (small sample, single setting, no peer review) and should be downweighted
- Any pattern across the papers that the research questions did not anticipatePrompt 3: Sinteza nalaza po temama preko izvora
I have a structured source log from a literature review on [tema] for [kontekst projekta]. I need to move from a list of source summaries to a synthesis organized by theme. Below is the extracted log.
Source log:
[ubacite strukturisane redove iz Prompta 2 — citat, nalaz, relevantnost itd.]
Please:
1. Identify 4-7 themes that recur across the sources. For each theme, name it in 2-4 words and write a one-line definition.
2. For each theme, write a one-paragraph synthesis that combines the evidence from multiple sources. Cite the sources in-text. Distinguish convergent findings (multiple sources agree) from contradictions (sources disagree) and from single-source claims.
3. For each theme, assess the strength of the evidence using these levels:
- Strong: 3+ peer-reviewed sources converge with consistent methods
- Moderate: 2-3 sources converge but methods or contexts differ
- Weak: 1-2 sources, single context, or contradictory
4. Flag any theme where the convergent finding contradicts conventional wisdom or stakeholder assumptions, and explain why.
5. Identify any source that consistently appears across multiple themes — it may be a foundational paper worth reading in full.
6. List the themes in order of relevance to my decision, not in order of evidence strength.Prompt 4: Gap analysis i prioritizovane preporuke
I have a thematic synthesis of a literature review on [tema] for [kontekst projekta]. The team needs to decide [konkretna dizajn ili istraživačka odluka] in the next sprint.
Synthesis (themes with strength of evidence):
[ubacite teme iz Prompta 3]
Original research questions:
[ubacite 2-4 pitanja sa početka pregleda]
Please:
1. Gap analysis: For each original research question, state explicitly whether the literature answers it (Strong / Moderate / Weak / Not at all), and what specific aspect of the question is unanswered.
2. List the gaps in priority order based on how much they block the team's decision. The top 2-3 gaps are candidates for primary research.
3. Recommendations: For each theme with Strong or Moderate evidence, draft a concrete recommendation tied to the team's specific decision. Each recommendation should include the finding, the source(s) behind it, the strength of the evidence, and the suggested action.
4. Identify any recommendation that contradicts the team's likely instinct or stakeholder assumptions, and frame it carefully (acknowledge the contradiction, present the evidence, suggest a small validation step).
5. Identify any decision where the literature is too thin to support a recommendation, and propose the smallest primary study that would close the gap.
6. Draft a 5-sentence executive summary the lead can use as the opening of the readout brief, anchored on the headline finding and the top recommendation.