GPT for Sheets and Docs

Accepts natural language descriptions of desired spreadsheet calculations and generates, fixes, or explains Google Sheets formulas (including QUERY, ARRAYFORMULA, VLOOKUP, etc.) by parsing user intent and mapping it to formula syntax. The extension reads the active spreadsheet structure to understand column names and data types, then uses the selected LLM provider to synthesize formulas contextually. Users can request formula creation, debugging of broken formulas, or explanations of existing formula logic without manual syntax lookup.

Applies data transformation rules across multiple rows in parallel by accepting natural language descriptions of cleanup operations (e.g., 'remove extra whitespace', 'standardize phone number format', 'fix capitalization') and executing them row-by-row using the selected LLM. The extension reads the target column(s), applies the transformation prompt to each row independently, and writes results back to the spreadsheet. Supports deduplication, validation, and normalization workflows without requiring formula knowledge or custom scripts.

Provides enterprise-grade security and compliance capabilities including Zero Data Retention (ZDR) policy ensuring data is not used for LLM model training, encryption in transit and at rest, Single Sign-On (SSO) via Google OIDC, and ISO 27001 certification. Supports BYOK (Bring Your Own Key) for organizations requiring private API endpoints or on-premise deployments, and GDPR compliance for EU data residency requirements. Enables enterprises to use AI automation while maintaining data privacy and regulatory compliance.

Generates or rewrites text content in bulk by applying a natural language prompt to each row of a spreadsheet column, with results written to a new or existing column. The extension sends each row's content to the selected LLM provider with the user's instruction (e.g., 'write a marketing email for this product', 'summarize this article in 50 words', 'translate to Spanish'), collects responses, and batches writes back to the sheet. Supports one-answer-per-row workflows for content creation, summarization, translation, and copywriting at scale.

Automatically assigns categories, tags, or classifications to rows of unstructured text by sending each row to the selected LLM with a classification prompt (e.g., 'categorize this customer feedback as bug, feature request, or complaint'), collecting the LLM's response, and writing results to a new column. Supports multi-label tagging, sentiment analysis, intent classification, and custom taxonomy assignment without requiring training data or machine learning expertise.

Augments spreadsheet rows with additional information by sending each row's content to the selected LLM with an enrichment prompt (e.g., 'look up the headquarters location for this company', 'find the founding year and industry'), collecting responses, and writing results to new columns. Supports web-aware LLM models (e.g., Perplexity, OpenAI with browsing) to fetch real-time information, or uses LLM knowledge cutoff for historical data. Enables non-technical users to add context, metadata, or derived fields at scale without manual research or API integration.

Processes images referenced in spreadsheet rows by sending image URLs or embedded images to vision-capable LLM models (e.g., OpenAI GPT-4V, Google Gemini, Anthropic Claude) with a natural language analysis prompt, collecting descriptions or extracted data, and writing results to new columns. Supports object detection, text extraction (OCR), quality assessment, and custom image analysis without requiring separate computer vision tools or expertise.

Translates or localizes text content across multiple rows by sending each row to the selected LLM with a translation prompt (e.g., 'translate to Spanish', 'localize for Japanese market'), collecting translated results, and writing them to new columns. Supports multiple target languages, tone/style preservation, and context-aware localization (e.g., adapting idioms or cultural references) without requiring professional translation services or language expertise.

Provides a chat-based interface in the Google Sheets sidebar where users can ask questions about the active spreadsheet, request insights, or get help with spreadsheet tasks through natural language conversation. The extension maintains context of the current spreadsheet and allows multi-turn conversations without requiring users to manually copy/paste data or switch between applications. Supports exploratory analysis, troubleshooting, and ad-hoc queries without predefined workflows.

Allows users to select and switch between multiple LLM providers (OpenAI, Anthropic, Google, Mistral, Perplexity, Parallel) at the operation level, with support for BYOK (Bring Your Own Key) configuration for enterprise users. The extension abstracts the underlying API differences and allows users to choose providers based on cost, capability, or latency requirements without changing their workflow. Supports provider-specific features (e.g., web search in Perplexity, vision in GPT-4V) through a unified interface.

Implements a prepaid credit model where teams purchase credit packs that are shared across all team members and operations, with transparent usage tracking per operation type (formula generation, bulk processing, etc.). Credits are consumed based on LLM provider pricing and operation complexity, and users can monitor remaining credits and usage history within the extension. Supports per-team credit pools without per-seat licensing, enabling flexible scaling and cost control.