A conversational spreadsheets agent that automates data collection, analysis, formula creation and decision-making via natural-language chat.
Key features
Conversational Interface: Interact with spreadsheets via natural-language chat to query data, request calculations, and instruct actions without writing formulas or scripts.
Data Collection Automation: Collects and consolidates inputs into spreadsheets through conversational prompts and connectors, reducing manual import and cleanup work.
Accurate Formula & Calculation Generation: Generates and inserts spreadsheet formulas and computed columns from user intents, aiming to reduce formula errors and improve correctness.
Decision Support & Recommendations: Analyzes sheet data to surface insights, recommendations, and suggested next actions to support decision-making directly from the sheet context.
NPi Action Library Integration: Provides an open-source action library (NPi) and developer APIs that let agents register and call external functions or tools to extend spreadsheet workflows.
Developer Extensibility: Enables developers to build custom functions, connectors, and tool integrations so agents can perform actions outside the spreadsheet and bring results back into sheets.
Natural-language spreadsheet queries and commands
Automated data collection from websites and sources
Structured spreadsheet generation in real time
Workflow automation for repetitive spreadsheet tasks
Focus on accuracy and error reduction
Conversational spreadsheet interface allowing natural-language commands to read, write, and analyze sheets
NPi action library: declarative tool/function API to expose actions to agents
Multi-language support and SDKs (primary languages: Python and Go; Starlark present)
Open-source distribution under Apache-2.0 with GitHub releases, issues and contribution workflow
Support for function decorators/@function to register callable tools
Stateful function-calling and mechanisms to save function-calling state and handle human confirmation
Extensible tool integrations enabling agents to operate external apps and pipelines
Release and versioning via GitHub (releases, commits, issue tracking)
Best for
Natural-language Formula Creation: Ask the agent to compute complex formulas or transformations and have it generate and insert correct spreadsheet formulas automatically.
Automated Data Intake: Consolidate survey, form, or external data into a single sheet by instructing the agent to fetch, normalize, and append records via conversational commands.
Report and Dashboard Generation: Convert raw data into summarized reports and visualizations by requesting the agent to aggregate metrics, create pivot tables, and prepare dashboards.
Repetitive Task Automation: Automate recurring spreadsheet workflows such as reconciliation, cleansing, or template population through agent-run actions and scheduled instructions.
Augmenting Sheets with External Tools: Use the NPi action library to let agents call external APIs or services (e.g., enrichment, validation) and write results back into the spreadsheet.
Decision Workflow Enablement: Run scenario analysis and get prescriptive recommendations (hiring, pricing, prioritization) by asking the agent to evaluate sheet data and propose actions.
Automating web data extraction into spreadsheets
Business reporting and decision support
Data analysis and cleaning without coding
Student and research data collection
Streamlining repetitive spreadsheet workflows for teams
Automate spreadsheet edits, calculations, and reporting via natural-language chat
Build agents that integrate spreadsheets with external services and enterprise workflows
Automate ETL, data validation, and aggregation tasks inside spreadsheets
Prototype and deploy custom tool functions for agent-driven automation using Python or Go
Enable human-in-the-loop workflows where agents request confirmations or save state during operations
