Obsidian Copilot vs Smart Ecosystem

A clear, no-fluff comparison of Copilot for Obsidian (a community plugin) and Smart Environment + Smart Plugins, focused on features, privacy/data flow, setup, and cost.

Disclosure: This comparison is published by Smart Connections. We link to primary sources and update this page when pricing, policies, or features change.

Last verified: 2026-02-17

paywalled supporter-only early-release supporter-only experimental missing or problematic

Features

Obsidian Copilot versus Smart Ecosystem feature comparison
Feature / Aspect	Obsidian Copilot	Smart Environment + Smart Plugins
Chat with notes	RAG PDFs Images	RAG (Notes & excerpts) PDFs Images Drag‑n‑drop files into context Context builder interface Named contexts
Recommended content	"Relevant Notes" sidebar	Smart Connections Block‑level (excerpts) connections Inline connections
Embedding models	Select an embedding provider/model (OpenAI, Google, Cohere, etc.) Local embeddings supported via local models like Ollama (requires setup) Plus: copilot-plus-small + copilot-plus-multilingual embeddings	Zero setup local embeddings (default bge-micro-v2) Stored on-device in your vault (.smart-env/) Configure embeddings/models in Smart Environment settings
Inline autocomplete	Ghost‑text autocomplete	—
ChatGPT Integration	—	Smart ChatGPT Smart Connect App
Safe AI editing	Overwrites inline	Smart Editor
Context engineering	Backlink scrape only	Smart Context
Templates & scaffolds	—	Smart Templates
Agents / tool calls	Experimental multi‑tool agents	Smart Actions Smart Chat MCP integration
Visual dashboards	—	Community contributed plugins: Smart Visualizer Smart Vault Visualizer
Chat models	Custom configuration for any OpenAI-style API	Custom configuration for any OpenAI-style API Easy setup for most popular platforms: Ollama LM Studio Open Router OpenAI Anthropic Claude Google Gemini Groq Azure OpenAI Grok (xAI)
Data storage	Orama DB	Dependency-free, easily-accessible JSON
Runtime dependencies	60+ third-party packages	1 third-party package (transformers.js)
Plugin bundle	3.5 MB (too big for AI audit, >1 MM tokens) Minified (unreadable)	1.2 MB (easy audit with AI, ~330 K tokens) Un‑minified (easy to read & change)
Data privacy	Vault QA uses a local data store (local index) Exclusions: prevent folders/notes from being indexed or sent Copilot Plus: requests processed through Copilot backend (even with your own API key)	Indexing and retrieval happen on-device by default Smart Environment data stored in your vault (.smart-env/) Cloud models are optional: you decide what context leaves your machine
Modularity	Monolithic plugin	Mix‑and‑match plugins Re-use embeddings for all Smart Plugins
Adaptability & Extendibility		Designed to be adapted for many platforms Easy for developers to build Smart Plugins
Install complexity	Requires installing third‑party software to keep data local	Zero‑setup local embeddings
Community & roadmap influence	Creator‑led roadmap Discord	Public GitHub Lean Coffee meet‑ups every other week Community supporter private chat
Cost	Free ($0) + bring your own API key (provider fees may apply) Plus: $14.99/month or $139.99/year (billed yearly) Self-Host Supporter: $349.99 one-time (includes 2 years of Plus) Believer plan is grandfathered (existing users only)	Core plugins: $0 Pro plugins (all-access): $30/month or $299/year (14-day free trial) Source available (see license)

Why dependency count matters

More third-party code can increase audit scope and supply-chain exposure. That can be totally reasonable for a feature-rich plugin, but it does make security review harder. When you care about privacy, treat "how much code do I need to trust?" as a real decision factor.

Smart Plugins aim to keep the core runtime surface small and readable. For any plugin, you should still review update practices, permissions, and the specific data flow you have configured.

Own your notes, own your models

Smart Plugins are local‑first. Swap Ollama, LM Studio, or a tiny transformer without changing a single note.

Build workflows, not vendor lock‑in

The Smart Ecosystem core exports plain JSON. Fork it, script it, audit it – no hidden API calls.

Community momentum

Solo founder means rapid feedback loops. Ideas from GitHub discussions ship in days, not quarters.

Privacy and data flow: what leaves your device

Privacy is configuration. The practical questions are (1) where embeddings run, (2) where the index/vectors are stored, and (3) whether requests are routed through an intermediate backend depending on plan.

1) Text transmission: embeddings and chat

Retrieval workflows (RAG / Vault QA) require turning note text into embeddings. If embeddings are computed locally, your note text does not need to be sent over the network to generate vectors. If embeddings are computed via a cloud API, the text being embedded is sent to that provider.

Copilot supports multiple embedding providers and also supports local models like Ollama (with setup). It also supports exclusions so you can prevent specific folders, tags, or note patterns from ever being indexed or sent. Copilot settings

Smart Environment defaults to on-device embeddings (no setup) and stores its data inside your vault. Smart Environment settings

2) Index and vectors: local file vs remote database

After embeddings exist, they must be stored somewhere for fast retrieval. Copilot's pricing page describes a local data store for Vault QA. Copilot pricing

Smart Environment stores its workspace in a .smart-env/ folder in your vault, and Smart Connections emphasizes that indexing and retrieval happen on-device by default. Smart Connections

The key distinction is not "vectors vs no vectors" - it is whether your index is stored locally and whether any provider receives your raw text during embedding or chat calls.

3) Routing, telemetry, and retention: plan and policy matter

Even if you bring your own API key, some plans may route requests through a backend for orchestration, stability, or telemetry. Copilot Plus states that requests are processed through their backend even when you use your own API key. Copilot Plus privacy policy

If you want the tightest privacy posture, aim for: local embeddings + local chat models + explicit exclusions for sensitive folders. Treat "offline" as something you verify (network monitor), not just a marketing claim.

Quick checklist

Choose where embeddings run: local (on-device) vs cloud (API).
Choose where chat runs: local model server vs cloud API.
Exclude folders/notes you never want indexed or sent.
Confirm where data is stored and how to delete/rebuild it.

Ready for private AI that listens to you, not a cloud dashboard?

Install Smart Plugins

One install, three plugins: Connections, Chat, Context.

Runs side‑by‑side with Copilot – try Smart Plugins risk‑free.

Frequently Asked Questions

Does Smart Connections work without internet?

Yes. Built-in embedding models are fully local, and can run offline. Chat models can also be run offline with the right setup.

Can I mix Smart Plugins with Copilot?

Absolutely. Smart Plugins respect standard file paths and generally run side-by-side with Copilot for Obsidian (community plugin). If two plugins compete for the same hotkeys or UI, disable the overlapping feature in one of them.

How do I get early features?

Become a Community Supporter to vote on and test new releases first.

Does Smart Ecosystem require extra installations?

No for local embeddings: a small on-device embedding model ships out of the box. If you want fully local chat (offline), you may need to run a local model server (for example Ollama or LM Studio) and point Smart Environment to it.