Features
Everything Keldura does
Turn the sources you follow into a living AI Library — then skim it, ask it, keep it current, and publish it.
Capabilities
| Feature | What it is | How it helps |
|---|---|---|
| Multi-source connect | Bring in YouTube, Instagram, websites, local video and audio, Google Drive documents, images, PDFs, Word docs, PowerPoint files, text, and Telegram via the Telegram Import Bot. | Bring in the sources you already follow when you want a personal Library, instead of converting everything first. |
| Video channel and playlist import | Paste a YouTube playlist or channel URL to transcribe each video into one searchable folder. | Build a searchable corpus from a whole video source in one job. |
| Import auto-sync | Keep supported YouTube channel, playlist, and website imports current on a schedule so new items join the Library over time. | Libraries stay up to date without re-pasting the same source URL for every update. |
| Video visual-text extraction | Sample video frames and extract visible text from slides, whiteboards, and diagrams with OCR. | Search and ask about information that was shown on screen, not only spoken aloud. |
| Website import | Crawl a domain or sitemap and turn every page into searchable text; enable auto-sync on supported website imports to pick up changes over time. | Turn docs, blogs, and knowledge bases into material your team can keep asking about as they update. |
| Local file and folder upload | Drag-and-drop audio, video, images, PDFs, Word documents, text files, or a whole folder. | Bring legacy recordings and documents into the same Library. |
| Library collections | Organize connected material into folders with counts, search, and quick actions. | Give every source a durable home before asking questions or sharing access. |
| Cited Chat | Ask questions across one or more Library folders and inspect citations from the source material. | Get answers you can verify instead of ungrounded summaries. |
| Suggested questions | Use source-aware prompts to start useful analysis quickly. | Move from connect to insight without designing every question from scratch. |
| Briefings | Turn updates from monitored sources into scheduled, source-linked digests; how many sources and digests you get depends on your plan. | Keep up with changing sources without reopening every item manually. |
| Keldura Daily | Read published, source-linked Daily cheatsheets without an account or personal sources. | Start reading immediately, then choose interests and add personal depth. |
| Reusable Bots | Define a Bot once with folders, prompts, and model settings. | Reuse the same knowledge behavior and prepare it for publishing where eligible. |
| Telegram Import | Forward or send messages and files to the Telegram Import Bot, or bind a group so new messages land in a Library folder (per-file size limits apply). | Bring chat and shared files into the same cited Library without re-uploading in the browser. |
| Bot channels | Connect an eligible Bot to a website widget or Telegram so people can ask the Library where they already work. | Publish one reusable Bot through multiple channels without duplicating its folders, prompts, or model. |
| MCP agent access | Expose Keldura knowledge to MCP-capable tools on eligible plans. | Let AI tools work with your folders on plans where this access is available. |
| Developer API | Use API keys and REST docs for folders, prompts, documents, chats, messages, and limits on eligible plans. | Build scripts, dashboards, and product workflows on top of Keldura data. |
| Saved prompts | Keep reusable prompt libraries and apply them in folder chat; share prompt groups with teammates. | Run the same analyses consistently across teams and collections. |
| Translation and summaries | Translate transcripts and generate summaries for connected material. | Skim and understand long sources faster. |
| Layered document extraction | Use extraction engines and OCR for native documents, scans, and images. | Make messy PDFs and visual documents searchable. |
| Sharing controls | Share folders and prompt groups with read or write access by account, link, or token, with revocation. | Collaborate on Libraries and instructions without sending raw files around. |
| Home dashboard | Resume recent Libraries, see source activity, and start from suggested questions. | Return to active knowledge work without hunting through folders. |
| ZIP export | Download a folder and its direct subfolders as a ZIP of plain-text documents. | Keep your knowledge portable and auditable. |
| Credit transparency | Review credit balance and model/source costs where available. | Understand usage as a transparent budget before running more work. |
| Progress streaming | Watch import and chat progress as work happens. | Know what is running without refreshing or guessing. |
| Soft-delete with undo | Use a short undo window on supported deletes. | Recover from accidental cleanup actions. |
| Free plan | Start on the free plan before choosing a paid plan. | Evaluate your own Library before paying. |
Popular workflows
How Keldura compares
Keldura vs. NotebookLM
Compare cited Libraries, recurring-source sync, sharing, and notebook research outputs.
Read the NotebookLM comparisonKeldura vs. ChatGPT Projects
Compare source-backed Libraries with a general-purpose project workspace inside ChatGPT.
Read the ChatGPT Projects comparisonKeldura vs. Chatbase
Compare direct Library work and reusable Bots with customer-support agent operations.
Read the Chatbase comparisonKeldura, from kelda
Kelda is Old Norse for a spring or well: a source that keeps flowing. The K-Spring mark pairs a well shaft with outward ripples, because in a living knowledge base, everything flows from the source.
Curious how it is built? Read how we ship Keldura with AI agents.