28308a14d6
* doc: split blog index into Hindsight and Hindsight Cloud sections - Tag the document upload post with `hindsight-cloud` - BlogListPage renders two sections, capping Cloud at 3 posts with a "View all →" link - Swizzle BlogTagsPostsPage so /blog/tags/hindsight-cloud uses the custom grid layout * doc: attribute blog posts to Nicolò Boschi with GitHub profile image Replace the generic "Hindsight Team" author with the real author entry (nicoloboschi) across all 15 blog posts. GitHub profile image is loaded from https://github.com/nicoloboschi.png. * doc: add Hindsight Team title to nicoloboschi author * doc: assign blog posts to correct authors based on git blame - Add benfrank241 (Ben Bartholomew) and chrislatimer (Chris Latimer) to authors.yml - Assign 7 posts to Ben, 1 post to Chris, remainder stay with Nicolò
2.7 KiB
| title | authors | date | tags | hide_table_of_contents | ||||
|---|---|---|---|---|---|---|---|---|
| What's New in Hindsight Cloud: Document File Upload |
|
2026-03-09 |
|
true |
Hindsight Cloud now lets you upload files directly to any memory bank. PDFs, Word documents, PowerPoint presentations, Excel spreadsheets, images, and plain text files are all supported.
Two Extraction Methods
When you upload a file, you choose how Hindsight processes it:
Standard extraction uses Markitdown to pull text from your document. It is free and works well for text-heavy files like PDFs, Word docs, and plain text.
Enhanced extraction (Iris) uses AI-powered processing to understand the structure and meaning of your document. This is useful for complex layouts, scanned documents, and images. Enhanced extraction incurs per-token charges.
Both methods extract the content and store it as structured memories in your bank, ready for recall and reflect queries.
How It Works
- Open any memory bank in Hindsight Cloud
- Click the upload button and select your files
- Choose Standard or Enhanced (Iris) extraction
- The Document Operations panel tracks progress with status indicators
Once processing completes, the extracted content becomes part of your memory bank. Your agents can recall and reflect on the document contents just like any other memory.
Supported File Types
- PDFs -- reports, whitepapers, research papers
- Word documents (.docx) -- meeting notes, specs, proposals
- PowerPoint presentations (.pptx) -- slide decks, training materials
- Excel spreadsheets (.xlsx) -- data tables, financial reports
- Images (.png, .jpg) -- screenshots, diagrams (Enhanced extraction recommended)
- Plain text (.txt, .md) -- logs, notes, documentation
What Else Is New
This update also follows two other recent additions:
Bank-scoped API keys (March 3) let you restrict API keys to specific memory banks. This is useful for multi-tenant setups where each customer or agent should only access its own memories. Unauthorized access attempts return a 403 error.
MCP support (February 13) added Model Context Protocol integration for AI clients like Claude, Cursor, and VS Code. You can connect in single-bank mode for a dedicated agent memory or multi-bank mode for cross-bank operations, with 30 tools available out of the box.
Get Started
File upload is available now in Hindsight Cloud. Upload your first document and try a recall query against it to see the results.