> For the complete documentation index, see [llms.txt](https://docs.diaflow.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.diaflow.io/changelog/2026/february-13-2026.md).

# February 13, 2026

Optical character recognition for documents, temporary public share links for AI chat, and comprehensive Google Workspace automation nodes.

See what’s new and improved in our latest update.

### Doc2Text OCR Engine for Scanned Documents

* The `Doc2Text` node now includes built-in Optical Character Recognition (OCR), enabling text extraction from scanned PDFs, images, DOCX files, and HTML web pages.
* Preserves tabular layout, heading hierarchies, and paragraph formatting for downstream LLM processing.
* Eliminates manual copy-pasting and third-party OCR subscription costs for invoice, contract, and receipt pipelines.

### 24-Hour Public Share Links for Agentic Chat

* Generate secure, temporary shareable URLs for any AI Agent chat session with a single click.
* Share links remain active for exactly 24 hours, allowing external stakeholders, clients, and partners to view research summaries and analyses without logging in.
* Sessions are read-only and redact private workspace credentials, ensuring absolute data security.

### Comprehensive Google Nodes Integration

* Built-in workflow nodes for Google Workspace: Google Docs, Google Drive, and Google Sheets.
* Automatically generate branded Google Docs from templates, injecting dynamic workflow variables and customer data.
* Read, upload, and organize team files directly within automated Google Drive folder hierarchies.

<details>

<summary>Improved</summary>

* **Resilient Network Retries:** Added automatic retry interceptors on the web client to recover from transient 502/503 gateway drops without losing unsaved changes.
* **Loop Output Normalization:** Standardized all loop iteration outputs with underscore keys (e.g., `item_index`, `item_value`), resolving edge-case input parser bugs.
* **WebScraper Audio Output:** WebScraper nodes can now detect and download embedded audio streams for downstream speech-to-text processing.
* **Numeric Input in Agentic Chat:** Handled pure numeric prompts gracefully, preventing agent classification errors when analyzing raw mathematical inputs.

</details>

<details>

<summary>Fixed</summary>

* **Google Drive Multi-File Upload:** Fixed an asynchronous promise race condition that caused batch uploads of multiple files to Google Drive to fail halfway through.
* **Google Docs Template Insertion:** Resolved a bug in the Google Docs node where creating documents from predefined templates threw 500 errors.
* **Published App Image-to-Video:** Fixed an issue where published standalone applications failed to convert uploaded images into AI-generated videos.
* **Cloudfront Presigned URLs:** Fixed an authentication signature mismatch causing direct file uploads via presigned URLs to be rejected.
* **Google API Key Image Creation:** Corrected credential scoping so personal Google API keys can be used for image generation without permission errors.

</details>