What's New in ResearchBuddy AI
See the evidence behind every claim — the Argument Heat Map is here, plus a sharper reading experience across the app.
Argument Heat Map — See How Strong Your Evidence Really Is
Every literature review now comes with an Argument Heat Map. Two AI models (GPT + Claude) independently analyse your sources, then we cross-validate: each claim is scored by real citation data — not the AI grading itself. You instantly see which arguments are rock-solid, which need more sources, and where the research community disagrees.
How to use it
Agent Mode — 6-Agent Research Pipeline
Submit a research question and watch 6 specialised AI agents work in parallel — decomposing your query, searching peer-reviewed papers, scraping the web, analysing, writing, and fact-checking — all automatically.
How to use it
Research Workspace
A unified project space where you can run all research tools — Literature Review, Paper Audit, Peer Review, Reference Check — inside a single organised session, with results saved to one project.
How to use it
Recent Updates
Bigger Chat Fonts & Better Loading Experience
April 2026Chat text is now larger and easier to read (16px body, improved heading hierarchy). The loading state has been completely redesigned — a pulsing card with a live progress bar and phase descriptions so you always know what ResearchBuddy is doing while it works.
Auto-Focus on Every Tool Page
April 2026When you open Peer Review, Prompt Coach, Paper Audit, or Upload Draft, the input area is now automatically focused and scrolled into view — no extra clicks to get started.
Domain-Aware Deep Audit
February 2026The Audit engine now detects your paper's domain and applies the right methodology rubric automatically. Econometrics papers get checked for OLS assumptions, heteroskedasticity, unit roots, and cointegration. Medical papers follow CONSORT/PRISMA. ML papers are validated against SOTA benchmarks and ablation requirements.
Triple-Model Cross-Validation
February 2026Deep Audit now runs three independent models — OpenAI o4-mini, Claude Sonnet 4.6, and Gemini 2.5 Flash — in parallel, then averages their scores. A finding only makes it into your report if multiple models agree. Fewer false positives, higher confidence.
Real Reference Verification
January 2026References are now verified against CrossRef and Semantic Scholar APIs in real time. You get the DOI, journal, citation count, and publication year for every paper cited — not just a title the AI made up.
Full Dark / Light Theme
January 2026The entire app now respects your theme preference consistently — including all inner pages (Audit, Peer Review, Uploads, History). Toggle from your profile in the sidebar.
AI Peer Review
December 2025Upload your paper and get structured reviewer feedback — Abstract clarity, Introduction framing, Methodology rigour, Results interpretation, and Discussion depth — formatted as a standard journal peer review.