Row-Bot
AI AssistantPython / NiceGUIApache-2.0

Row-Bot Review

A local-first desktop AI assistant for reasoning, tool orchestration, and durable memory.

Deployability
2/5
Value
4/5
Privacy
4/5

Each review covers deployability, value versus commercial alternatives, and privacy model. Tools that can run locally were started and exercised; mobile or backend-dependent tools were assessed from published builds, source code, and deploy guides. Ratings reflect what we were able to verify.

Row-Bot — Open Source Local-First AI Assistant

A local-first desktop AI assistant for reasoning, tool orchestration, and durable memory. Last updated: 2026-06-21.

One-sentence verdict: A feature-rich, privacy-oriented AI assistant with installers for Windows, macOS, and Linux; running from source is heavy and was not completed in the documented deploy environment.


What the System Is

Row-Bot (formerly Thoth) is a local-first desktop AI assistant built in Python with a NiceGUI frontend. The deploy guide documents the intended setup but reports the service was not started because the dependency set is large (PyTorch, transformers, sentence-transformers, LangGraph, FAISS, optional local models, and more).

The README highlights:

  • Reason / Orchestrate / Work agent loop via LangGraph ReAct.
  • Durable memory — personal knowledge graph, semantic/lexical/graph recall, Obsidian-compatible export.
  • 30+ core tools — web search, documents, browser automation, Gmail, calendar, shell, workflows, channels, image/video generation, Wolfram Alpha, and more.
  • Developer Studio — Git workspace linking, code threads, diffs, commits/PR prep, optional Docker sandbox.
  • Designer Studio — decks, documents, mockups, exports to PDF/HTML/PPTX.
  • Workflows and real-time voice — scheduled runs, webhooks, Telegram/Discord/Slack/SMS, voice sessions.
  • Provider-aware routing — Ollama local models plus OpenAI, Anthropic, Google, xAI, OpenRouter, Atlas Cloud, MiniMax, and custom OpenAI-compatible endpoints.
Key data
Category AI Assistant
Language Python / NiceGUI
License Apache-2.0
Self-hosted Yes
AI provider Ollama, OpenAI, Anthropic, Google, xAI, OpenRouter, Atlas Cloud, MiniMax, or custom endpoints
Desktop installers Windows, macOS, Linux
Stars (GitHub) 1,284

How to Install and Deploy

The project recommends the published installers rather than running from source:

  • Windows — download the installer from GitHub Releases.
  • macOS — download the DMG and drag Row-Bot.app to Applications.
  • Linux — run the one-line installer:
curl -fsSL https://raw.githubusercontent.com/siddsachar/row-bot/main/installer/install-linux.sh | bash

The documented source path from the deploy guide is:

cd /data2/docker/going-global/repos/row-bot
python -m venv .venv
source .venv/bin/activate
python -m ensurepip --upgrade
python -m pip install -r requirements.txt
python app.py

The deploy guide notes that full dependency installation requires significant disk space and download time, and that some features (voice, desktop notifications, system tray, browser automation) need additional system libraries that may not be available in a headless environment.


How to Test

The documented test flow is:

  1. Install Row-Bot via the platform installer or source path.
  2. Launch the app and complete the setup wizard (Local, Providers, or Custom/Self-hosted).
  3. Verify chat loads and accepts a prompt such as Show my headache trends this month.
  4. Add a provider key and confirm a hosted model responds.
  5. Test memory by asking the assistant to remember a fact, then recall it in a new thread.

The repository also includes smoke and integration scripts under scripts/ and tests/.


Privacy & Compliance

Row-Bot is not HIPAA compliant. Its privacy model is stronger than most cloud-only assistants: durable app data stays local by default, there is no Row-Bot account or telemetry pipeline, and provider keys are stored in the OS credential store when available. However, any chat sent to a hosted provider (OpenAI, Anthropic, etc.) leaves the machine. Health data should only be used with local Ollama models and still lacks HIPAA safeguards.


Row-Bot vs Commercial AI Assistants

Dimension Row-Bot Commercial Assistants (e.g., ChatGPT, Claude, Copilot)
Cost Free / self-hosted Subscription / freemium
Data model Local memory and files; provider calls optional Cloud-hosted conversations
Offline use Possible with Ollama local models No
Tool breadth 30+ core tools, MCP, plugins, workflows Closed, vendor-curated
Setup effort Heavy from source; installers available Sign-in in browser/app
Mobile / web access Desktop-focused Native apps and web

Who Should Use It

  • Power users who want a local-first assistant with durable memory and broad tool support.
  • Developers building agentic workflows, custom tools, or MCP integrations.
  • Privacy-conscious users willing to run local models via Ollama.

Who Shouldn't Use It

  • Users who need a lightweight, instantly runnable open-source app.
  • Teams looking for a HIPAA-compliant health tool.
  • Users who primarily need mobile or web access.

FAQ

Is Row-Bot a health-specific tool?

No. It is a general-purpose AI assistant. The README includes health-related example prompts, but it is not a medical device or clinical system.

Can it run without sending data to the cloud?

Yes, if you configure a local model through Ollama and avoid cloud tools/embeddings. The setup wizard explicitly offers a Local mode.

Was the service started in the deploy environment?

No. The deploy guide reports that dependencies were too heavy to install and start in the headless environment; it recommends using the published installers or a machine with sufficient disk space and optional GPU.


Verdict

Row-Bot is the most ambitious general-purpose assistant in this batch. Its local-first design, broad tool set, and installer distribution make it a genuine alternative to cloud-only assistants for desktop power users. The source path is not beginner-friendly, and it is not a health compliance tool, but the published installers lower the barrier significantly.

Ratings: Deployability 2/5 · Value vs Commercial 4/5 · Privacy Compliance 4/5