DEVONtechnologies LLC

DEVONthink 4.3.2 is here. Now supports Claude Fable 5 (when it becomes available once more) and Ollama image diffusion models, plus a new LM Studio connector. Also: faster MCP search and improved redaction before sending data to AI. We fixed a crash on the macOS 27 Developer Beta, too. #devonthink #ai #claude #fable #ollama #lmstudio #macos27

DEVONthink lets you use several common file formats: PDF documents for reading, rich text and Markdown for writing, and many more. But if you need to record information in a spreadsheet-style format, we also have a native format for that: sheets. A DEVONthink sheet is a file format for displaying tabular data in an accessible and editable way. If you have ever used Microsoft Excel or Apple Numbers, sheets will look familiar to you. Composed of user-definable columns and rows, it is useful for keeping records on a variety of topics, like genealogy, software licenses, customer information, bibliographic data, and more. You can find some useful example sheets in DEVONthink’s Data > New from Template menus, e.g., Education > Books. Create a sheet via the Data > New menu. Define the columns you need by giving each a name and choosing a data type, e.g., text, date and time, URL, country, a Boolean (checkbox) option, and more. Or if you have a TSV (tab-separated values) or CSV (comma-separated values) file, you can import and use it as a sheet. Work on a sheet by clicking a cell and entering the information. Tab to the next column, or press ⏎ to create a new record or jump to the next row. You can edit the columns, add or delete rows via the Tools > Sheets commands or the buttons in the Editing bar above the view/edit pane. When viewing a sheet, you can click the column headers to sort on that column. For Pro and Server users, you can use the View > Document Display > Form View to view each record as a form, a view familiar to FileMaker users. One thing to be aware of: despite the similar appearance to Excel or Numbers, sheets are plain text under the hood and don’t support advanced features like formulae or conditional formatting. #devonthink #tsv #csv

DEVONthink To Go 4.1.3 is out. It gives the integrated help pages a visual overhaul and resolves a number of issues, including a crash in the chat, UI delays, and deadlocks. It also fixes unformatted dates in feed articles and syncing bundle documents. #devonthinktogo #update

In the past, we have covered how to scan to DEVONthink To Go using a third-party app and how to use your iPhone as a scanner for DEVONthink. But you can also use the built-in scan feature of DEVONthink To Go to scan directly into the current database, and even make searchable, multi-page documents. First, set up OCR. Go into Settings > OCR and enable Make scans searchable, choose a language if needed, and adjust further options to suit your situation. Now go into a database, tap the Add button (+) and choose Scan. The first time you do a scan, you’ll be prompted to allow use of the camera. Once the permission is granted, position the mobile device over a document. You’ll see a blue, dynamic polygon, moving as it tries to detect page edges. Once it determines where it thinks the page is, it will automatically capture the page and a thumbnail of the captured page appears at the lower left. To switch to manual capture instead, tap the Shutter button. If one scan is all you need, tap the checkmark button and the document will be saved in the database, and OCR will make it searchable. If you are scanning a multi-page document, change the page and continue the process until finished. Then tap the checkmark button to save the scans into one document. To quickly access the scanner from your home screen, long-press DEVONthink To Go’s application icon and choose New Scan. The app will open and you can follow the process described above. #devonthink #devonthinktogo #capture #scanning

DEVONthink 4.3.1 further refines our Markdown parser with options like better image styling and captions, improved inline footnotes, and some fixes. This release also improves the MCP server, including safer handling of custom metadata, setting item aliases, opening documents externally, and using OpenLibrary as a fallback for Google Books. Additionally, we updated some image-generation AI models and now support Claude Opus 4.8. #devonthink #markdown #mcp #ai #claude https://buff.ly/iUTcx90

Very soon, macOS will no longer support Intel Macs. Upgrading hardware is also a good impetus to upgrade DEVONthink. If you’re running the Pro or Pro Office editions of the 2.x line, here is the smoothest migration path to version 4. At no point should you run any uninstaller application or try to manually uninstall DEVONthink 2 or 3 until instructed to below. Also, make sure you have the last release of version 2 installed: 2.11.3. If needed, you can download it and replace your current version from our legacy downloads. These instructions are written so you can follow them without guesswork. It will go much faster than you imagine! - In the Finder, press ⇧⌘A to open the Applications directory. Leave this window open. - In your web browser, download the latest version of the 3.x line from our Legacy Downloads. - Unzip the archive, open the disk image, and install DEVONthink 3 into the Applications directory. - Launch DEVONthink 3 and quit it. - Back in the Applications directory, move the DEVONthink 3 and DEVONthink Pro / Pro Office applications to the trash. - In your web browser, download the latest version of 4.x from our Downloads. - Unzip, open the disk image, and install it in the Applications directory. - Launch and quit DEVONthink 4. - Restart your Mac. - After the machine has restarted, empty the trash. - Read the Getting Started > First and Last Steps section in the built-in Help and manual. Your global inbox should be intact and open. Now just do a Spotlight search for kind:database name:dtBase2 and open your databases. If you were running DEVONthink 2.x Personal, contact our customer support and we’ll help you with the transition. #devonthink #macos #upgrade #migration

DEVONthink To Go 4.1.2 is available! This maintenance release adds emergency notification support so we can reach you quickly if a critical update is held up for some reason. We've also fixed a whole range of issues, including ones related to item counting, wrong RSS article dates, and more. #devonthinktogo

MCP (Model Context Protocol) is a mechanism that allows AI assistants to communicate with apps to accomplish tasks for you. On macOS Sequoia and later, DEVONthink now has its own MCP server built-in, presenting new opportunities to access and work with your databases from outside the application. First, set up DEVONthink’s MCP server in Settings > AI > MCP by enabling the server for one of the built-in providers: Anthropic Claude, OpenAI Codex, or the Hermes Agent. Then switch to that application and try describing what you want, e.g., Find all my notes about project X or Create a summary document in my Research database's inbox. The AI checks the available commands and carries out the task, including: reading and summarizing records, creating and updating documents, moving items between groups, and adding tags or annotations, etc. The more specific and thorough the prompt, the better the result. Information provided via MCP responses, like the content of invoices you’ve requested to be examined and graphed, is sent to the AI provider’s servers online. But DEVONthink reinforces data privacy by redacting or obscuring personal information like credit card numbers, email addresses, etc. if you’ve enabled Redact sensitive content in the MCP settings. Also, the AI acts on whichever databases are open at the time. If you don’t want the external AI to “see” an item, enable Exclude from Chat & MCP in the Generic Info inspector. For databases, enable the same option in the File > Database Properties or merely close that particular database. #devonthink #mcp #artificialintelligence #automation

DEVONthink 4.3 Herschel* is the third feature release for DEVONthink. It introduces its own MCP server, a bridge from your databases to compatible AI apps like Claude or Codex. Comprising almost 60 commands, we built it with data privacy and security at its core: you control local and remote access, and it redacts private information before it leaves your machine, honors items excluded from AI, and blocks direct filesystem access to your database internals. A menu bar icon appears whenever its HTTP server is running, giving you quick access to server controls, settings, and the URL you need to connect. And because some databases should simply never touch an AI, the Exclude from Chat & MCP option lets you exclude entire databases at once — enabled automatically for encrypted and revision-proof databases. Beyond the MCP server, DEVONthink 4.3 Herschel brings sweeping AI improvements. The chat assistant now redacts more sensitive information, e.g., phone numbers, credit card numbers, authentication tokens, and labeled secrets, before anything is sent to a language model, a protection that extends to AI smart actions and summarize and transform functions. It supports the Apple Intelligence Foundation model for simple tasks, and you can choose which AI model handles automatic tagging. We have also updated numerous AI models, including Claude Opus 4.7, GPT 5.5, Gemini 3.5 Flash, Mistral Small 4 and Medium 3.5, Flux 2, and many OpenRouter models. Replacing MultiMarkdown, our new Markdown parser and renderer supports callouts, citations, captioned tables, and improved CriticMarkup. It is also more compatible with AI-generated output. Markdown documents know about smart quotes and dashes, and converting them to plain text strips all formatting and renders MathJax in readable form. Finally, new desktop widgets for macOS Sequoia and later let you quickly view or access your Reading List or Favorites, load a workspace, or read the useful tips we regularly publish on our blog. For Server users, the web interface gets a dialog for opening provided item links and lets you reveal, move, duplicate, and replicate items from the context menu. The See Also inspector has a new section for classifying documents, and we have rebuilt the PDF viewer with a table of contents, page thumbnails, and full-text search across all browsers. * We name this release after the periodic comet 35P/Herschel–Rigollet, discovered in 1788 by Caroline Lucretia Herschel (1750–1848). Caroline was the sister of the famous astronomer William Herschel and made significant contributions to astronomy in her own right. She was the first known professional female astronomer, became the first woman to receive a salary as a professional astronomer, and received the Gold Medal of the Royal Astronomical Society. #devonthink #update #mcp #markdown #ai #claude #chatgpt #gemini #mistral

AI inside DEVONthink can be a powerful tool. But when it comes to AI, some users have security concerns, such as possible prompt injections. So, what exactly are they, and are they a risk in DEVONthink? A prompt injection happens when text hidden inside a document is crafted to circumvent AI with its own commands, like one that says, “Ignore the previous prompt and do this instead…” The danger isn’t limited to triggering unwanted commands — injections can also silently distort AI responses such as summaries or evaluations. In addition, injections can occur not only in your own documents but also in web search results and downloaded pages, which is why searches and downloads should be restricted and only enabled when needed. This isn’t a DEVONthink problem specifically but an issue for any document processing AI system. This is why AI models are increasingly designed to be more resistant to these attacks, and the best precaution is to be mindful about what you add to your database. While in a well-curated personal database the practical risk is low, there are features in DEVONthink to limit or thwart these injections: - AI responses are automatically stripped of unexpected or potentially dangerous HTML content (such as scripts or frames) that could result from prompt injections. - Even if you allow downloads, the Chat assistant, Chat smart actions, and AppleScript commands automatically disable downloading web pages while processing a document. The Chat assistant also disables downloads after accessing the content, annotations, or comments of a document. You can also limit what the AI is allowed to access, both inside and outside your databases. However, keep in mind that stricter limits may reduce the AI’s ability to assist you: - Enabling Search > Sources: Web in the settings or Search: Web in the Chat’s Options is often unnecessary and increases exposure to prompt injections via search results or downloaded pages. Enable it only as needed. - If you do enable web searching, consider also disabling the AI > Chat > Assistant: Allow download of web pages and Load remote images automatically options. Disabling remote images also prevents prompt injections from silently transferring data to remote servers. - You could disable AI > Search: Database and also AI > Chat > Assistant: Allow property & content changes. You can temporarily enable database searching in the Chat assistant as needed. - When using the Chat assistant or Chat-related smart actions, e.g., Chat - Query, set the Documents popup to Without Document, unless you are specifically dealing with documents. We don’t intend to cause undue concern, but to help you make better-informed choices in how you use AI. #devonthink #devonthinktogo #ai #artificialintelligence #security #promptinjection