Changelog ChatAcademia

Stay up to date with all the latest updates, improvements, and new features we ship to help you do better research. We’re constantly improving ChatAcademia for academics and researchers.

Home

© 2026 ChatAcademia

Study Steward, Save to Library & Better Document Reading

Study Steward, Save to Library & Better Document Reading

Study Steward helps researchers turn messy study operations into an actionable project plan, while agents can now read papers from pasted IDs or URLs and save resolved papers back to connected reference managers.

Improvements

  • Start the new Study Steward beta agent from the app sidebar and agent discovery surfaces
  • Map study workflows into stages, risks, owners, milestones, and handoff actions
  • Draft ethics and compliance checklists for consent, privacy, recruitment, vulnerable populations, and approvals
  • Build practical data management plans covering storage, access, retention, sharing, backups, and documentation
  • Plan study timelines and submission packets with progress, readiness, and blocker summaries shown beside the chat
  • Paste a DOI, PubMed ID, ArXiv ID, or supported paper URL so agents can read the paper content directly
  • Ask agents to save resolved papers to Zotero or Mendeley, including metadata, tags, optional folders or collections, and the resolved PDF when available
  • Find Papers now keeps searching and adding papers even when Zotero or Mendeley save tools are available in the same chat
  • Reconnect guidance is clearer when Zotero or Mendeley needs fresh permissions before saving papers
  • Dropbox, Google Drive, Zotero, and Mendeley file access is more reliable when tokens expire or files need provider-specific download handling
  • Sign-in and sign-up cards now use the intended desktop width again while preserving right-to-left layout support
  • Paste a public document URL so agents can read PDFs, Word files, spreadsheets, CSVs, images, text, markdown, or HTML pages without a manual upload
  • Document-reading tool credits are now attributed to the chat that used them so usage totals stay easier to audit
  • Memory tool activity now keeps a clear title even when a memory model or document response has an empty display name

Dropbox, Claim Evidence & Hallucination Checker Agents

Dropbox, Claim Evidence & Hallucination Checker Agents

ChatAcademia now adds two beta agents for checking claims against the literature and auditing bibliographies, alongside Dropbox storage support for bringing your research library into conversations.

Improvements

  • Connect Dropbox from the Storage section and use Dropbox files as chat attachments
  • Start the new Dropbox Agent to browse folders, search files, inspect metadata, and read document content
  • Use Dropbox files in supported project and extraction workflows without re-uploading them manually
  • See Dropbox beta status clearly in the agent selector, chat header, breadcrumb, and Storage sidebar entry
  • Use Claim Evidence to paste an argument or paper DOI, extract checkable claims, and compare supporting and refuting papers in a project panel
  • Use Hallucination Checker to paste references or DOIs, open links to matched real papers, and flag likely hallucinated, mismatched, duplicate, or updated entries
  • Review stance-labeled evidence tables, audit progress, confidence, and review queues directly beside the chat

Clearer Credit Guidance & File Upload Checks

Clearer Credit Guidance & File Upload Checks

The pricing page now makes it easier to understand how credits work before choosing a plan. Reading Assistant and Mock Peer Review also guide you to attach the required files before starting, avoiding confusing failure messages.

Improvements

  • Open the new How do credits work? link directly from the pricing section
  • Compare monthly credit allowances across Free, Basic, Elite, and Ultimate plans
  • Review a process-by-process credit table for each agent, including searches, paper additions, citation checks, verdicts, and document operations
  • Use the model reference table to understand how credit usage may vary across ChatAcademia Fast, Pro, and Max depending on task complexity
  • See a clear inline prompt when starting Reading Assistant or Mock Peer Review without the required PDF or manuscript file

Multilingual App, Reliable Chats & App Startup

Multilingual App, Reliable Chats & App Startup

ChatAcademia now supports a multilingual in-app experience, with browser-language detection, language switching, and localized interface copy across core research workflows. Chats, onboarding, PDF exports, and app startup are also more resilient in edge cases that previously caused internal errors or crashes.

Improvements

  • Choose your display language in the app, or let ChatAcademia start with your browser language automatically
  • Core app navigation, auth screens, account settings, agent routes, project panels, dialogs, buttons, tool renderers, dates, numbers, and empty states now use localized interface text
  • The language selector now supports a larger set of display languages with search, browser-language defaulting, and a layout that stays stable for longer language names
  • AI responses can receive the selected response language so generated research help stays aligned with the interface language
  • The ChatAcademia website now highlights multilingual app support on the main landing page with a searchable language directory
  • Chats and onboarding streams no longer fail when final response chunks arrive in an unexpected order
  • Memory setup during onboarding is more reliable when the backend is under database lock pressure
  • PDF exports now handle Greek letters and other scientific symbols by converting them to PDF-safe text instead of crashing
  • The app now handles browsers that block local storage more gracefully, reducing startup and settings-related failures

Smarter Errors & Guided Chat Recovery

Smarter Errors & Guided Chat Recovery

The app now routes known problems to the right surface — a focused modal, a clear toast, or inline text — instead of generic failures. When older chat history cannot be replayed by the current model, you get a dedicated explanation with a one-click path to start fresh in the same agent.

Improvements

  • Known server errors carry stable codes so the interface can show the right message and action (for example, insufficient credits opens the upgrade flow instead of a vague error)
  • Corrupted or incompatible chat history surfaces a dedicated modal that explains what happened and offers to open a new chat in the same workspace
  • Rate limits and permission issues use consistent, human-readable copy instead of opaque status text
  • Shared handling across agents and chats reduces mismatched error toasts between the sidebar, chat view, and project panels