Frequently Asked Questions

ResearchImpact AI is an automated research impact assessment platform designed for university researchers and research teams. It generates structured, evidence-based impact reports by analysing your publications and publicly available research outputs in minutes rather than days.

Each report follows the Research Impact Assessment Framework (RIAF), covering problem significance, research outputs, current impacts, and projected future impacts. Reports are delivered as ready-to-use Word documents, an executive summary, and a structured Excel workbook.

ResearchImpact AI is built for:

• Individual researchers preparing grant applications, performance reviews, or national research evaluation submissions
• Research team leaders documenting the collective impact of their group
• University departments and research offices compiling institutional impact evidence
• Grant writers and research administrators who need to articulate research value quickly and credibly

If you need to explain what your research has achieved and why it matters, ResearchImpact AI is for you.

A research impact assessment documents the real-world difference your research has made beyond academic publications. It answers questions like: What problem did this research address? What outcomes has it achieved in policy, practice, health, or society? What is the expected future benefit?

Research impact assessments are required for national evaluation exercises such as the UK's Research Excellence Framework (REF) and similar frameworks globally. They are also increasingly used in grant applications, institutional reporting, and stakeholder communications.

The Research Impact Assessment Framework (RIAF) is the four-section structure used to organise every ResearchImpact AI report. It is grounded in the NSW Health Research Impact Assessment Framework and has been adapted across multiple research domains:

1. Problem Significance — the challenge your research addresses and why it matters
2. Research Outputs — publications, tools, datasets, and other deliverables produced
3. Current Impacts — real-world outcomes achieved to date, with evidence
4. Future Impacts — projected long-term benefits with evidence-based rationale

This structure aligns with the requirements of major national research evaluation frameworks and provides a rigorous, consistent basis for impact narratives.

ResearchImpact AI combines automated literature search, web research, and AI-powered synthesis to build your report. When you submit a request, the system:

1. Searches academic databases and the web for publications and evidence related to your research topic and team
2. Analyses the gathered evidence to identify key themes, outcomes, and impact indicators
3. Generates a structured narrative across the four RIAF sections, with citations
4. Produces formatted output files ready for use or further editing

In Professional mode, you can also upload your own documents such as unpublished papers, policy briefs, or technical reports so the analysis draws on evidence not publicly available online.

Both modes generate a full research impact report covering the same four sections: problem significance, research outputs, current impacts, and future impacts.

Starter is designed for a quick first look at your research impact. It produces a concise report (~1,500–2,000 words) in around 10–15 minutes, ideal for exploring the tool or getting a rapid overview before a grant application or internal review.

Professional goes deeper. It searches more broadly, reasons more carefully about your research context, and produces a more comprehensive report (~2,500–3,000 words) in around 20–30 minutes. It also unlocks the ability to upload your own documents (papers, policy briefs, annual reports) so the analysis can draw on materials not publicly available online.

Use Starter for a fast, solid overview. Choose Professional when you need a richer, more thorough report for impact case studies, national research evaluation submissions, or external stakeholder communications.

Each report is focused on a single research topic. This is by design, as a tightly scoped topic produces a more coherent and useful impact narrative than a broad one. If your work spans multiple topics, we recommend running a separate report for each and combining the outputs.

Reports can be generated across ten domains:

• Agriculture & Food Security
• Climate & Environmental Science
• Digital Technology & AI
• Education & Learning Sciences
• Energy & Sustainability
• Engineering & Applied Technology
• Fundamental Sciences
• Healthcare & Medical Research
• Social Sciences & Economics
• Urban Planning & Architecture

Selecting the closest domain helps tailor the analysis framework and search strategy to your field.

ResearchImpact AI draws on publicly available research publications, web sources, and (in Professional mode) your own uploaded documents. The reports are designed to provide a well-structured, evidence-informed starting point for your impact narrative.

References are tracked throughout the pipeline, so every claim stays tied to the source it came from. Each report then goes through an automated reference check before delivery, with citations verified against the scholarly record, corrections applied and broken links flagged.

This does not remove the need for your own review. Verify key claims against their original sources before using the report in a formal submission, particularly specific statistics and attributed findings.

Yes. Once a report is complete, open it from your dashboard and choose Adapt for grant. Pick a target scheme and the platform rewrites your impact evidence to match that funder's criteria, language and structure, producing a separate Word document alongside your original report.

Your original report is never overwritten, and you can generate adaptations for several schemes from the same report.

Thirty commonly used schemes across five jurisdictions:

• Australia: ARC National Interest Test, Discovery Projects, DECRA, Future Fellowships, Linkage; NHMRC Investigator, Ideas, Synergy and Partnership grants
• United Kingdom: REF 2029 impact case studies, UKRI R4RI, Future Leaders Fellowships, Wellcome Career Development, Royal Society URF, NIHR Advanced Fellowship, Innovate UK KTP
• Europe: Horizon Europe Section 2 (Impact), ERC Starting, Consolidator and Advanced, ERC Synergy, MSCA Postdoctoral Fellowships
• United States: NSF Broader Impacts, NSF CAREER, NIH R01, K99/R00, R35 MIRA
• Canada: Tri-Agency Common CV, CIHR Project, SSHRC Insight and Partnership

Browse the full list, with eligibility rules, word limits and deadlines, in the Grant Scheme Guide. The list is not exhaustive. Always confirm requirements on the funder's own website before applying.

Yes. Every scheme carries its own format constraints and the adaptation is written against them. The ARC National Interest Test is produced to a strict 200-word plain-English limit, while a Horizon Europe impact section is structured quite differently.

Funder requirements change, so check the current guidelines before you submit.

It is optional, but it helps. If you provide a draft, or just a short summary of your aims and approach, the adaptation can speak to the proposal you are actually writing rather than to your past work alone. This particularly improves the forward-looking sections.

You can leave out confidential detail, or leave the field blank entirely, in which case the adaptation draws solely on the impact analysis already in your report. You can also add a short steer, such as "lead with the health-economic angle".

It is sent to our AI model provider solely to produce the adaptation. ResearchImpact does not store it: no database record, no logs, no audit copy. Our model provider's API terms exclude using your inputs to train their models. Only the generated document is saved, and only to your own dashboard.

If your institution restricts sharing pre-submission text, provide a high-level summary instead, or leave the field blank.

Check the scheme's own policy. That is the safe default. Several major funders have published guidance on generative AI in applications. Some require disclosure, some restrict particular uses, and the rules are still changing.

Whatever the policy, an adaptation is a drafting aid, not a submission. You are the author of your application: review it, verify every claim and citation, and rewrite it in your own voice before it goes near a submission portal.

Each report is focused on a single research topic and a defined group of researchers, which gives the analysis the depth and specificity needed to produce a meaningful impact narrative.

For larger teams or institutions, we recommend a group-by-group approach: divide your researchers into smaller core groups by research topic or department, run a separate report for each group, and then combine the outputs into a broader institutional report. This approach also tends to produce higher-quality results, since each report can focus on a coherent body of work.

Yes. This is our most common institutional request. Reports are customised in consultation with your institution and tailored to your specific requirements. Three layers can be adapted:

• Research template: what the analysis looks for. A template defines the evidence types, impact categories and search strategy appropriate to your field, programme or funder.
• Report structure: the sections, their order, and the length and emphasis of each, so output matches the format your institution already uses rather than ours.
• Document branding: your own Word template, logos and styling, so reports look like your institution's own documents.

We have already delivered this for a health research programme, whose reports use a bespoke research template, a tighter section structure and their own branded Word output. If you would like the same, get in touch.

Yes. Institutional deployments are built around the systems you already run. We connect research information systems such as Pure, Converis, Symplectic Elements, DSpace and EPrints, or a read-only link to your institutional data warehouse, so researcher profiles, publications and grants flow through without manual entry. Single sign-on through your existing identity provider (SAML, OAuth2 or Active Directory) can be included.

Every institution's environment is different, so we scope the connection with your team. Contact us to talk it through.

Yes. Institutional deployments include programmatic access, so reports can be submitted, tracked and retrieved directly from your own systems. Impact reporting becomes part of your existing workflows and reporting cycles rather than something run by hand.

API access is arranged as part of an institutional agreement rather than enabled on individual accounts. Contact us to discuss what you need.

The essentials are your research topic, a few keywords, the researchers involved, and a research domain.

The remaining inputs refine the analysis: the research period to cover, your organisation and country, the language style and variant (seven registers, British or American English), a free-text guidance field, and any documents you choose to upload.

It lets you steer the analysis in your own words rather than accept a generic reading of your topic. Use it to:

• point at impacts you already know about, such as a policy citation, a clinical guideline or a tool that has been adopted
• name collaborators, partner organisations or end users worth looking for
• exclude work that shares your topic but is not yours
• ask for emphasis on a particular audience or outcome

Short, specific instructions work better than long ones. For example: "Focus on adoption by state health departments rather than clinical trials."

This field is for brief instructions. If you have substantial material to bring in, such as reports, briefs or strategy documents, upload it using the document upload option instead.

Match it to what you are evidencing. A grant application or fellowship usually needs a defined recent window, such as the project itself or the last five years. A promotion case or national evaluation submission may need a decade or more.

Bear in mind that impact lags research. Policy uptake, guideline changes and citation of your work often appear years after publication, so a window that stops at your most recent paper may miss the outcomes you are trying to evidence. If you are unsure, start wider and narrow it on a second run.

For two reasons.

First, they help find the right publications. Author names are rarely unique, and a search on a common surname will return work by people who are not you. Knowing the institution and country lets the analysis separate your publications from those of researchers with similar names.

Second, impact is assessed in context. Your organisation and country tell the analysis where to look for the policy documents, funding bodies, regulatory and health systems, and local stakeholders most likely to show your research being used. The same topic will surface quite different evidence depending on where the work was based.

Yes. Document upload is available in Professional mode. You can upload up to 10 files in PDF or DOCX format, each up to 10MB and 100 pages.

This is the most direct way to bring in evidence the analysis cannot find on its own. Useful documents include:

• unpublished papers, theses and technical reports
• policy briefs, submissions and evaluation reports
• letters of support, testimonials and partner correspondence
• annual reports, project reports and internal impact summaries
• institutional, state or national strategy documents your research aligns to, so the report can set out that alignment explicitly

Uploaded documents are used solely to inform your report and are deleted from our servers once processing is complete.

Credits are the currency used to generate reports. Each submission costs:

• Starter report: 1 credit
• Professional report: 3 credits
• Grant adaptation: 2 credits

Credits are deducted at the time of submission. New accounts receive an initial credit allocation to get started.

In the rare event that a report fails to complete, the credits used for that submission are not automatically refunded. Please contact us at info@researchimpact.ai and we will arrange a replacement.

To purchase additional credits, please contact us at info@researchimpact.ai.

Your completed report is delivered as a ZIP file containing:

• Use_Case_Study.docx — the full impact report
• Executive_Summary.docx — a condensed summary suitable for presentations or grant applications
• Research_Impact_Data.xlsx — a structured workbook with key findings organised for institutional analysis or further reporting
• Markdown/ — plain-text versions of the report and individual sections, useful for editing or embedding in other documents
• input_settings.txt — a record of the parameters used to generate the report

Research_Impact_Data.xlsx provides a structured, machine-readable version of the report's key findings. It is particularly useful for teams running multiple reports who want to compare or aggregate results across research groups when compiling an institutional submission or building a research portfolio overview.

Reports cannot be re-run or edited directly through the platform once they are complete. If you need changes, the most practical approach is to edit the downloaded DOCX file directly, or submit a new report with revised inputs. If you need support with a specific report, contact us at info@researchimpact.ai.

You do. You retain all rights to your research and to the reports generated from it. Use a report in a grant application, an impact case study, a promotion dossier, an institutional submission or a public-facing document. That is what it is for.

Two things to keep in mind:

• You are the author. The report is a drafted starting point, not a finished submission. Review it, verify every reference against its original source, and rewrite it in your own voice before it goes out under your name.
• Check the receiving body's AI policy. Funders, publishers and evaluation exercises increasingly set their own rules on AI-assisted text, and those rules govern what you may submit.

We do not claim ownership of your research content, we do not publish or share your reports, and we do not use them to train AI models.

ResearchImpact AI is currently available to researchers at accredited universities and research institutions worldwide. Sign-up requires a valid institutional email address.

If your university or institution email is not recognised during sign-up, it may not yet be on our approved list. Please contact us at info@researchimpact.ai with your institution name and email domain and we will add it promptly.

Yes. All data is transmitted over encrypted connections. Uploaded documents are used only to generate your report and are deleted from our servers once processing is complete.

We only use large language models (LLMs) from trusted, established providers, including OpenAI, Google Gemini and Anthropic, accessed on paid enterprise APIs. Their terms exclude using customer inputs for training, and each publishes independent security and compliance documentation.

We do not store your research data beyond what is needed to deliver your report, we never share or sell it, and we do not use it to train models of our own.

Your data is processed on enterprise-grade cloud infrastructure in Australia. The large language models are external services and may process data outside Australia, always over encrypted connections and under enterprise API agreements whose terms exclude training on your inputs.

If your institution has specific data residency requirements, or you plan to submit highly sensitive or confidential material, contact us before submitting and we will work through your requirements with you.