SciSpace Agent Prompting Guide: Research Examples and Checks

SciSpace Agent is your AI Co-scientist, designed for researchers who want to streamline their everyday research work. Whether you’re finding the latest papers, drafting a manuscript, analyzing patents, or discovering suitable grants, SciSpace Agent helps you do all this in one workspace with a single prompt. 

But here’s the catch — the quality of the result (what you get) depends on how you prompt (ask). Just as a poorly designed experiment produces weak results, vague prompts produce shallow answers. This is why we have created this guide: it covers the principles of good prompting, shows you what makes a good prompt, and provides relevant examples grounded in real research scenarios. 

By the end of this guide, you will:

• Write prompts that are clear, structured, and unambiguous.
• Provide context and scope so the Agent “understands” your research intent.
• Refine and iterate prompts to sharpen results.
• Specify outputs in the exact format you need — tables, abstracts, literature maps, even LaTeX reports.
• Avoid common pitfalls like vagueness, information overload, or unclear objectives.

These guidelines are aimed at researchers and academics who use SciSpace Agent. They compile prompt-engineering best practices to help you get the most out of the agent. 

Before we dive into prompting techniques, let’s understand what SciSpace Agent is built to do.

SciSpace Agent isn’t just a chatbot — it’s an AI Co-Scientist designed to automate various research tasks:

1. Literature Research & Discovery Search across multiple databases: SciSpace, PubMed, arXiv, Google Scholar. Run systematic literature reviews. Map citation networks. Discover relevant journals and conferences.
2. Data Analysis & Processing Extract data from research papers. Perform statistical analysis and visualizations. Compare studies and run meta-analysis support. Synthesize findings across datasets.
3. Academic Writing & Communication Draft and format manuscripts. Generate reports in Markdown, LaTeX, or PDF. Build posters, presentations, or summaries. Even create research dissemination websites.
4. Research Tools & Workflows Identify grant opportunities. Search patents and clinical trials. Cross-validate references. Assist in proposal preparation.

When you give SciSpace Agent a prompt, it follows a structured process:

1. Query Analysis – Understands your research intent and constraints.
2. Planning – Strategizes execution (e.g., which databases to search, what filters to apply).
3. Multi-Source Research – Pulls insights from multiple databases and sources.
4. Synthesis – Combines information into a coherent narrative or dataset.
5. Output Generation – Produces structured deliverables: tables, reviews, abstracts, and grant sections.

This workflow is why clarity in your prompt is critical — it helps the Agent plan and execute without guesswork.

Most users assume the Agent will automatically understand their intent. But when prompts are vague, the Agent adapts broad behavior, by default:

• searching across multiple databases
• expanding the topic scope
• generating generic outputs

For example, a prompt like:

“Find papers on cancer”

It may retrieve results from multiple databases and create a broad synthesis even if your intention was to search only PubMed or focus on a specific cancer subtype.

This is the default Agent behavior, not an error. The more specific your instructions are, the more precise and research-ready the output becomes.

Example: #1 Vague prompt

Example: #2 Well-defined prompt

SciSpace Agent is designed to maximize relevance when instructions are incomplete, which means:

• It may search across all available databases
• Broaden the interpretation of your topic
• Choose generic output formats like markdown reports

This behavior is useful for exploration, but for research workflows requiring precision, prompts should include:

• The scope
• Output requirements
• Inclusion/Exclusion criteria
• Preferred databases

Structured prompting reduces ambiguity and gives the Agent a clearer execution path.

Getting good results out of an AI Agent isn’t about prompting whatever comes to mind; it’s about giving instructions that are precise, structured, and actionable. Follow the core prompting principles below:

Be as specific as you can. Broad, lazy prompts will only give you generic summaries that aren’t research-ready. For example:

❌ “Tell me about CRISPR.”

✅ “Analyze the effectiveness of CRISPR gene editing in treating sickle cell disease, focusing on clinical trials from 2020–2025.”

See the difference? The first one could pull the answer from anywhere; the second frames a clear task with a defined scope.

Always provide the background the Agent needs so it knows why you’re asking and what kind of answer you expect. That means including:

• Research objective: What exactly are you trying to achieve?
• Scope: How broad or narrow should the investigation be?
• Constraints: Any limits on time period, geography, or methodology?
• Output preferences: Do you want a summary, a table, a structured report, or something else?

Without context, you’ll just get filler. But, with the right context, you get something useful and reproducible.

Start wide if you’re exploring, then progressively sharpen your query until the output is exactly what you need.

• Initial: “Research machine learning in healthcare.”
• Refined: “Systematic review of deep learning applications in medical imaging for cancer detection, focusing on diagnostic accuracy studies from 2022–2025.”

That small shift from broad to refined turns a random overview into a literature review you can actually use in your work.

The verbs you use matter. Don’t just say “tell me” or “give me.” Use action words that guide the Agent into doing the right kind of work:

• Analyze → Deep dive into data or literature.
• Compare → Lay out differences side by side.
• Synthesize → Pull multiple sources into one cohesive answer.
• Evaluate → Judge quality, credibility, or impact.
• Summarize → Condense into a concise, structured overview.

The clearer the action, the more precise the response.

CORE is a mnemonic used in this guide to organize prompts; it is not a validated research method. Treat prompts as instructions for a workflow, and inspect sources, analysis choices and generated files before using the result.

A practical way to structure prompts is using the CORE framework.

CORE stands for:

Explain the background, objective, or research scenario.

Specify the format you want:

• report
• table
• PDF
• slides
• literature review
• citations

Assign a perspective to the Agent:

• peer reviewer
• researcher
• policy analyst
• grant evaluator

Define what should NOT be included:

• older studies
• non-peer-reviewed work
• irrelevant populations
• unwanted databases

This framework reduces ambiguity and significantly improves output quality.

If you want the SciSpace Agent to give useful, research-ready results, your prompts need structure. Always, direct the Agent with a blueprint: define the task, the boundaries, and the expected output. Otherwise, you’ll end up with scattered or generic results.

Start by telling the Agent exactly what you want it to do. Be concrete. Examples:

• “Conduct a comprehensive literature review on [TOPIC]”
• “Analyze the relationship between [VARIABLE A] and [VARIABLE B]”
• “Create a research poster summarizing [RESEARCH FINDINGS]”

The more specific the task, the more targeted the output.

Define the boundaries so the Agent knows what to include and what to leave out:

• Temporal: “Published between 2020–2025”
• Geographic: “Studies conducted in North America and Europe”
• Methodological: “Randomized controlled trials only”
• Population: “Adult participants aged 18–65”

Scope gives your research focus and prevents irrelevant information from cluttering the results.

Be explicit about how you want your results delivered:

• Format: “Generate a LaTeX report with figures and tables”
• Length: “10-page comprehensive review”
• Style: “Academic writing suitable for journal submission”
• Citations: “APA format with complete reference list”

This ensures you get a ready-to-use output instead of raw text you have to clean up later.

Set standards for the sources and studies the Agent should use:

• Source requirements: “Peer-reviewed journals only”
• Impact factor: “Journals with IF > 3.0”
• Study quality: “High-quality studies with large sample sizes”

Quality criteria make your results credible and research-grade.

Here’s a simple, reusable template you can plug into every query:

[TASK TYPE]: [SPECIFIC OBJECTIVE]

Context:
- Research question: [YOUR QUESTION]
- Background: [RELEVANT CONTEXT]
- Motivation: [WHY THIS MATTERS]

Requirements:
- Scope: [BOUNDARIES AND LIMITATIONS]
- Sources: [PREFERRED DATABASES/TYPES]
- Time frame: [PUBLICATION DATES]
- Quality criteria: [STANDARDS FOR INCLUSION]

Output:
- Format: [DESIRED FORMAT]
- Length: [APPROXIMATE LENGTH]
- Style: [WRITING STYLE]
- Citations: [CITATION FORMAT]

Additional considerations:
- [SPECIAL REQUIREMENTS]
- [CONSTRAINTS OR PREFERENCES]

Different types of research tasks require slightly different prompting approaches. Here’s how to structure prompts for common research workflows.

When you want a thorough, reproducible review of the literature, a systematic approach is the key. Example:

Task: Conduct a systematic review on the effectiveness of mindfulness-based interventions for anxiety disorders.

Search parameters:
- Databases: PubMed, PsycINFO, Google Scholar
- Time frame: 2018–2025
- Study types: Randomized controlled trials (RCTs) and meta-analyses
- Population: Adults with diagnosed anxiety disorders
- Intervention: Mindfulness-based stress reduction (MBSR) or mindfulness-based cognitive therapy (MBCT)
- Outcomes: Anxiety symptom reduction measured with validated scales

Output requirements:
- PRISMA-compliant methodology
- Risk of bias assessment
- Forest plots for meta-analysis
- 15-page LaTeX report with executive summary
- Complete reference list in APA format

Quality criteria:
- Peer-reviewed publications only
- Minimum sample size of 30 participants
- Follow-up period of at least 8 weeks

For exploring trends, historical evolution, or conceptual developments, a narrative review works best. Example:

Task: Create a narrative review on the evolution of artificial intelligence in drug discovery.

Focus areas:
- Historical development (2010–2025)
- Key technological breakthroughs
- Current applications and case studies
- Challenges and limitations
- Future directions and opportunities

Structure:
- Introduction with problem statement
- Methodological approaches in AI-driven drug discovery
- Success stories and breakthrough drugs
- Regulatory and ethical considerations
- Industry adoption and investment trends
- Conclusions and recommendations

Output: 20-page academic review suitable for Nature Reviews Drug Discovery, with high-quality figures and comprehensive citations.

When comparing interventions, technologies, or datasets, the key is to define metrics and subgroups clearly. Example:

Task: Compare COVID-19 vaccine efficacy across platforms (mRNA, viral vector, protein subunit).

Analysis framework:
- Efficacy metrics: Infection prevention, severe disease, hospitalization, death
- Safety profiles: Adverse events, contraindications
- Population subgroups: Age, comorbidities, immunocompromised status
- Variant effectiveness: Alpha, Delta, Omicron
- Durability: Effectiveness over time, booster requirements

Data sources:
- Phase III clinical trial data
- Real-world effectiveness studies
- Regulatory agency reports (FDA, EMA, WHO)
- Peer-reviewed publications (2020–2025)

Deliverables:
- Statistical analysis with confidence intervals
- Comparative tables and forest plots
- Risk-benefit analysis
- Policy recommendations
- Interactive dashboard for visualization

For funding research, prompts should define your target agencies, project scope, and output expectations. Example:

Task: Identify and analyze federal funding opportunities for climate change adaptation in coastal communities.

Search criteria:
- Funding agencies: NSF, NOAA, EPA, DOE
- Award amounts: $500K–$5M
- Duration: 3–5 years
- Focus areas: Sea-level rise, coastal erosion, ecosystem adaptation
- Collaboration requirements: Interdisciplinary teams preferred

Analysis components:
- Funding trends and priorities
- Success rates and competition levels
- Application requirements and deadlines
- Alignment with current research priorities
- Potential collaboration opportunities

Output:
- Comprehensive funding landscape report
- Ranked list of opportunities with fit scores
- Application timeline and strategy recommendations
- Template for proposal development

Sometimes a single prompt isn’t enough. Complex research questions often require layered approaches. Here’s how to make the most of SciSpace Agent with advanced techniques.

Break down large or complex tasks into manageable phases. It should follow the path — exploration → analysis → synthesis.

Stage 1: Broad Exploration

• Task: Conduct an initial scoping review on renewable energy storage technologies.
• Goal: Map the current research landscape, identify major categories, and highlight emerging trends.

Stage 2: Deep Dive

• Task: Based on Phase 1 findings, perform a detailed analysis of the three most promising storage technologies.
• Focus: Technical specifications, cost analysis, scalability, and commercial readiness.

Stage 3: Synthesis

• Task: Combine insights from Phases 1–2 into a comprehensive technology roadmap.
• Deliverables: Investment recommendations, policy implications, and a clear action plan.

Breaking tasks into stages keeps the Agent focused and ensures outputs are thorough and actionable.

Sometimes you need the Agent to adjust based on available data or research conditions. This is conditional prompting.

Example: Analyze the impact of remote work on employee productivity.

• Primary focus: Quantitative studies with productivity metrics
• Secondary analysis: If quantitative data is limited, include high-quality qualitative studies
• Tertiary consideration: If research is sparse, expand to hybrid work models

This approach allows flexibility without compromising scientific rigor.

Refine your prompts step by step to narrow down outputs to exactly what you need.

Initial request: “Research the relationship between social media use and mental health.”

Refinement 1: “Focus specifically on adolescents aged 13–18.”
Refinement 2: “Emphasize longitudinal studies showing causal relationships.”
Refinement 3: “Include intervention studies testing social media reduction strategies.”

Final synthesis: Integrate findings into evidence-based recommendations for parents and policymakers.

Iterative refinement ensures the Agent delivers research-grade, highly targeted results rather than broad, generic summaries.

Different research domains have unique requirements. Clear, domain-specific prompts help the SciSpace Agent deliver precise and actionable results. Here’s how:

Task: Investigate the molecular mechanisms of Alzheimer's disease progression.

Biological focus:
- Amyloid-beta plaque formation and clearance
- Tau protein hyperphosphorylation and neurofibrillary tangles
- Neuroinflammation and microglial activation
- Synaptic dysfunction and neuronal loss

Methodological preferences:
- Animal model studies (transgenic mice)
- Human post-mortem brain tissue analysis
- Neuroimaging studies (PET, MRI)
- Biomarker research (CSF, blood-based)

Temporal scope: Research published 2020–2025 with breakthrough findings.

Output: Mechanistic pathway diagrams with detailed molecular interactions.

Task: Analyze advances in quantum computing hardware architectures for practical applications.

Technical specifications:
Qubit technologies: Superconducting, trapped ion, photonic, topological
Error correction methods and fault-tolerance thresholds
Scalability challenges and solutions
Decoherence mitigation strategies

Application domains:
Cryptography and cybersecurity
Drug discovery and molecular simulation
Financial modeling and optimization
Machine learning and AI acceleration

Industry analysis:
Major players and approaches (IBM, Google, Rigetti, IonQ)
Investment trends and market projections
Regulatory considerations and standardization efforts

Deliverable: Technical roadmap with timeline predictions and investment recommendations.

Task: Examine the societal impact of universal basic income (UBI) pilot programs worldwide.

Analytical framework:
Economic outcomes: Employment effects, income distribution, poverty reduction
Social outcomes: Health, education, social cohesion, crime rates
Behavioral changes: Work patterns, entrepreneurship, skill development
Implementation challenges: Funding mechanisms, political feasibility, and targeting

Geographic scope:
Completed pilots: Finland, Kenya, Stockton (CA), Barcelona
Ongoing programs: Various cities and regions globally
Comparative analysis across different economic contexts

Methodological rigor:
Randomized controlled trials preferred
Quasi-experimental designs accepted
Long-term follow-up studies (>2 years)
Control for confounding variables

Output: Policy brief with evidence-based recommendations for implementation.

Even experienced users can fall into common prompting traps. Recognizing these mistakes and knowing how to fix them ensures the Agent produces usable, high-quality outputs.

Problem: “Research cancer treatments”
Solution: Be specific about treatment type, patient population, and timeframe.
Example: “Analyze the efficacy and safety of CAR-T cell therapy for relapsed/refractory B-cell acute lymphoblastic leukemia in pediatric patients, focusing on FDA-approved treatments and clinical trials from 2020–2025.”

Problem: “Find papers about machine learning”
Solution: Provide context about the application area, objectives, and industry relevance.
Example: “Conduct a literature review on machine learning applications in predictive maintenance for manufacturing equipment, focusing on reducing downtime and maintenance costs in the automotive and aerospace industries.”

Problem: “Analyze all research on climate change”
Solution: Narrow the focus to a specific region, crop, or timeframe for actionable results.
Example: “Systematic review of climate change impact on agricultural productivity in Sub-Saharan Africa, focusing on crop yield studies for maize, wheat, and rice from 2015–2025.”

Problem: “Write a report on renewable energy”
Solution: Define length, format, and specific content deliverables.
Example: “Create a 25-page technical report on offshore wind energy potential in the North Sea, including cost analysis, environmental impact assessment, and policy recommendations, formatted for submission to energy policy journals.”

Problem: “Find information about vaccines”
Solution: Set clear standards for source quality, study design, and publication type.
Example: “Systematic review of mRNA vaccine technology, including only peer-reviewed studies from high-impact journals (IF >5), focusing on mechanism of action, efficacy data, and safety profiles from Phase III trials.”

Assist in preparing a research manuscript for submission to Nature Medicine on our findings regarding CRISPR-Cas9 gene editing for treating beta-thalassemia.

Manuscript components needed:
- Literature review for the Introduction section (focus on current gene therapy approaches)
- Methods section review for clarity and completeness
- Results interpretation and statistical analysis verification
- Discussion section development with clinical implications
- Figure and table optimization for journal requirements

Specific requirements:
- Word limit: 3,000 words (excluding references)
- Reference limit: 50 citations (high-impact journals preferred)
- Figure requirements: Maximum 4 figures, publication-quality
- Ethical considerations: Patient consent and regulatory approval documentation

Additional support:
- Identify potential peer reviewers in the field
- Suggest appropriate supplementary materials

Develop a comprehensive research proposal for an NIH R01 grant application investigating the role of gut microbiome in Parkinson's disease progression.

Proposal sections:
- Specific Aims (1 page): Clear, testable hypotheses
- Research Strategy (12 pages): Significance, Innovation, Approach
- Preliminary Studies: Supporting data and pilot results
- Research Team: Expertise and collaboration plan
- Budget Justification: Personnel, equipment, supplies

Scientific focus:
- Microbiome-gut-brain axis mechanisms
- Longitudinal cohort study design (n=500, 5-year follow-up)
- Multi-omics approach (metagenomics, metabolomics, proteomics)
- Clinical correlations with motor and non-motor symptoms

Compliance requirements:
- NIH formatting guidelines
- Human subjects protection protocols
- Data sharing and management plans
- Rigor and reproducibility considerations

Conduct a meta-analysis examining the effectiveness of telemedicine interventions for managing chronic diseases during the COVID-19 pandemic.

Inclusion criteria:
- Study design: Randomized controlled trials and quasi-experimental studies
- Population: Adults with chronic conditions (diabetes, hypertension, heart disease)
- Intervention: Telemedicine platforms (video consultations, remote monitoring, mobile apps)
- Comparison: Standard in-person care or usual care
- Outcomes: Clinical indicators, patient satisfaction, healthcare utilization
- Time period: March 2020 - December 2024

Statistical analysis:
- Random-effects model for pooled estimates
- Subgroup analysis by disease type and intervention modality
- Heterogeneity assessment (I² statistic)
- Publication bias evaluation (funnel plots, Egger's test)
- Sensitivity analysis excluding high-risk studies

Deliverables:
- PRISMA flow diagram
- Forest plots for primary outcomes
- Summary of findings table
- Risk of bias assessment using Cochrane RoB 2 tool
- Manuscript draft suitable for Journal of Medical Internet Research

Perform a comprehensive patent landscape analysis for artificial intelligence applications in drug discovery and development.

Patent search parameters:
- Patent databases: USPTO, EPO, WIPO, Google Patents
- Time frame: 2018-2025
- Technology areas: Machine learning, deep learning, AI algorithms
- Application domains: Drug target identification, lead optimization, clinical trial design
- Geographic scope: US, Europe, Japan, China

Analysis components:
- Patent filing trends and growth patterns
- Key assignees and inventors (companies, universities, individuals)
- Technology clustering and classification
- Citation analysis and patent families
- Freedom to operate assessment
- Competitive intelligence and white space identification

Visualization and reporting:
- Patent landscape maps
- Technology evolution timelines
- Competitive positioning charts
- Investment and licensing opportunity analysis
- Strategic recommendations for R&D planning

Output: Executive summary with actionable insights for pharmaceutical companies and biotech startups

Getting better results from the SciSpace Agent isn’t just about writing a prompt — it’s about designing your search and outputs smartly.

1. Leverage Multi-Database Searches: Always specify multiple relevant databases to ensure coverage. Example: “Search across SciSpace, PubMed, arXiv, and Google Scholar for comprehensive coverage.”
2. Use Temporal Constraints Strategically

• Recent research: 2022–2025 for the latest findings
• Historical perspective: 2010–2025 for trend analysis
• Breakthrough periods: 2018–2025 for CRISPR studies

1. Specify Output Formats Early: Define exactly how you want your results presented. Example: Generate results as:

1. Include Quality Filters: Set standards to prioritize high-quality sources. Example: “Prioritize peer-reviewed publications, high-impact journals (IF >5), studies with large sample sizes (n >100), and systematic reviews or meta-analyses.”
2. Plan for Iterative Refinement: Start broad, then narrow down based on initial findings. Example: “Initial broad search, followed by focused analysis of the top 3 most relevant themes identified.”

Even with well-structured prompts, issues can arise. Here’s how to fix common problems:

Issue: Results are too broad
Solution: Add constraints

• Narrow the time frame
• Specify methodology
• Define population parameters
• Add geographic limitations

Issue: Insufficient results
Solution: Expand your search

• Broaden the inclusion criteria
• Extend the time range
• Include related terms and synonyms
• Consider grey literature sources

Issue: Poor quality sources
Solution: Strengthen quality criteria

• Specify journal requirements
• Add methodological filters
• Set minimum impact factor thresholds
• Exclude predatory journals

Issue: Inconsistent results
Solution: Standardize methodology

• Follow systematic review protocols
• Apply consistent inclusion/exclusion criteria
• Request a bias assessment
• Specify data extraction procedures

Issue: Overwhelming information
Solution: Structure outputs

• Request tiered summaries
• Ask for prioritized findings
• Specify key outcome measures
• Request concise executive summaries

In scientific research, clarity is everything. We write precise hypotheses, follow rigorous methods, and report data in standard formats so that others can replicate and build on our research work. Prompting follows the same logic; the clearer and more structured your instructions are, the more accurate and useful the Agent’s output will be. Let’s look at the example below of a good and a vague prompt! 

A vague prompt like:

❌ “Tell me about climate change research.”

It will trigger an equally vague answer — a generic overview you could find in any news article. But if you refine it into something like:

✅ “Summarize peer-reviewed articles published between 2020–2025 on machine learning applications for predicting climate change impacts. Focus on datasets used, methods compared, and reported accuracy. Present findings in a table with citations.”

You’ll suddenly see the Agent acting less like a chatbot and more like an actual Co-Scientist or a domain-aware assistant.

Prompting is just not about “talking to an AI.” It’s about structuring your research questions in a way that leverages the Agent’s strengths, including speed, scale, and synthesis.

A well-structured prompt helps the SciSpace Agent understand your task, tailor its response, and produce more useful results. Keep these practices in mind when writing your prompts:

• Be clear and specific
  Describe the task precisely and include key details such as the topic, audience, style, and length. Specify whether the content is for general readers, students, or domain experts, and whether you want a formal academic tone, comprehensive summary, or technical report.
• Provide context and scope
  Give the agent relevant background and explain the research setting. For example, mention whether the task is “for a biology journal” or “using the attached data.” This helps the agent tailor its response to your needs.
• Use examples or templates
  When possible, provide a sample output, such as a previous abstract or your preferred format. Examples give the agent a clearer idea of the structure and style you want and can make the response more relevant.
• Encourage step-by-step reasoning
  For complex queries, ask the agent to break down its approach. Phrases such as “explain your reasoning” or “show the steps” can help produce more detailed and thorough answers.
• Break down complex tasks
  If your request has several parts, divide it into stages or ask for an outline first. Smaller steps can make complex tasks easier to understand, review, and refine.
• Specify the output format
  Tell the agent how you want the response structured, such as in bullet points, a table, or sections. For example, ask it to “answer in bullet points” or “provide a short paragraph summary.”
• Iterate and refine
  Use the agent’s response to improve your prompt. If something is unclear or incomplete, ask a follow-up such as “elaborate on the step ‘Screening titles and abstracts’” or “shift the tone to be more academic.” Small adjustments can improve the final result.
• Maintain academic integrity
  Ask the agent to consider potential bias in cited studies, preserve originality when paraphrasing, and be transparent about limitations in the data and sources. This helps keep the output credible and aligned with academic standards.

• Use precise keywords related to your research area.
• Request structured outputs (tables, outlines, summaries).
• Validate outputs by cross-checking with original sources.
• Keep prompts within one clear task.
• Experiment with prompt rephrasing to improve results.

• Don’t rely on a single prompt for complex tasks (e.g., full paper draft).
• Don’t expect the Agent to replace peer-reviewed research.
• Don’t overload prompts with multiple unrelated tasks.
• Don’t assume citations are always correct without verification.
• Don’t use casual or vague wording like “Tell me something interesting.”

If your task needs a reproducible search, specify databases, search date, eligibility rules and the evidence table you want returned. For a systematic review, consult PRISMA-S when documenting the search; a prompt alone is not a review protocol.

• Will I verify sources before using them in academic work?
• Am I prepared to iterate if the first output is not perfect?
• Have I limited my prompt to one main task?
• Did I specify the format I want the output in?
• Did I provide enough context (field, timeframe, scope)?
• Have I clearly defined my task (review, draft, summarize, search)?

SciSpace Agent is more than just an AI tool; as mentioned, it’s your AI Co-Scientist, which is built to reduce the load of manual research and accelerate your research. But, as with any AI, the clarity of prompting determines the quality of output. By following these guidelines, you’ll get to explore the full potential of SciSpace Agent, making your research journey not just faster but also more effective and reliable. Try it now!