Glossary
Key terms and concepts used throughout Batcher.ai.
AI Assistant — The conversational copilot built into Batcher.ai that helps create workflows through natural language dialogue. See AI Assistant.
API key — A unique authentication token provided by an AI provider (OpenAI, Anthropic, Google, etc.) that allows Batcher.ai to send requests on your behalf. You add API keys when creating configurations.
Batch processing — Executing AI formulas across all rows of a spreadsheet in one operation, rather than processing data one row at a time.
Cell reference — A reference to a specific cell in the spreadsheet, such as A2 or B10. Used in formulas to dynamically pull data from your spreadsheet.
CONCAT — A function that joins multiple text strings and cell references into one value. Commonly used to build dynamic prompts: =CONCAT("Translate: ", A2).
Configuration — A saved set of parameters (provider, model, API key, temperature, etc.) that defines how an AI model is accessed. Referenced by name in formulas: =LLM(A2, "my-config").
Default model — The free, built-in AI model included with every Batcher.ai account. Referenced as "default" in formulas.
Formula — An expression starting with = that computes a value. In Batcher.ai, formulas can call AI models (=LLM()), combine text (=CONCAT()), or perform standard spreadsheet calculations.
GEO (Generative Engine Optimization) — Optimizing content for AI-powered search engines and generative results, similar to SEO but targeting AI-generated answers.
Haloscan — A specialized integration for SEO keyword research. Returns search volumes, related keywords, and SERP data as JSON. Referenced as "haloscan" in formulas.
Jina — A specialized integration for web scraping. Converts any URL into clean markdown content. Referenced as "jina" in formulas.
Level-by-level processing — A processing mode that respects column dependencies. If column C depends on column B results, all of column B is processed first.
LLM (Large Language Model) — An AI model trained on text data that can generate, analyze, and transform text. Examples include GPT-4, Claude, and Gemini.
LLM() — The core Batcher.ai function that sends a prompt to an AI model and returns the response: =LLM(prompt, "config_name").
Max tokens — The maximum number of tokens (roughly words) that an AI model will generate in its response. Higher values allow longer responses but cost more.
Prompt — The text instruction sent to an AI model. A good prompt clearly describes what output is expected. In Batcher.ai, prompts can include static text and dynamic cell references.
Provider — A company or service that hosts AI models (e.g., OpenAI, Anthropic, Google). Each provider requires its own API key.
Restore point — An automatic snapshot of your spreadsheet state saved before each write operation. Allows undoing changes.
SEO (Search Engine Optimization) — Optimizing content to rank higher in search engine results. A common use case for Batcher.ai workflows.
Sequential processing — A processing mode where cells are executed one at a time, row by row, without considering column dependencies.
Spreadsheet — The core workspace in Batcher.ai where data is stored, formulas are written, and AI processing happens. Similar to an Excel or Google Sheets document.
Streaming — Real-time delivery of AI responses, where text appears progressively as the model generates it rather than waiting for the complete response.
System prompt — A fixed instruction set in a configuration that is prepended to every AI request. Useful for consistent behavior (e.g., “Always respond in French”).
Target range — The range of cells where a formula should be applied (e.g., C2:C150). The AI Assistant uses this to expand a single formula template to all rows.
Temperature — A parameter that controls the randomness/creativity of AI responses. Low values (0-0.3) produce focused, deterministic output. High values (0.7-1.0) produce more varied, creative output.
Token — The basic unit of text that AI models process. Roughly equivalent to 3/4 of a word. Both input (your prompt) and output (the response) consume tokens.
Undo — Reverting the spreadsheet to a previous state using an automatically saved restore point. Multiple undo levels are supported.
Vision — The ability of certain AI models to analyze image content. When enabled, Batcher.ai automatically detects image URLs in prompts and sends them as visual input.
Workflow — A complete data processing pipeline built in a spreadsheet, consisting of input data, one or more formula columns, and output results. Workflows range from simple (one formula) to multi-step pipelines.