LangChain
LangChain · Writing · Productivity
The Executive AI Assistant is an open-source AI email agent from LangChain that triages your Gmail, drafts replies in your voice, and books meetings while you stay in control.
About LangChain
What Is the Executive AI Assistant
The Executive AI Assistant, or EAIA, is an open-source agent built by LangChain to take over the repetitive parts of managing email. It watches a Gmail inbox you connect, sorts every new message, writes draft replies that match your tone, and checks your calendar when someone asks for a meeting. You review what it produces in a shared inbox before anything goes out.
The project lives in the langchain-ai/executive-ai-assistant repository on GitHub and ships under the MIT license. Under the hood it runs on LangGraph, LangChain's framework for stateful agents, and it uses models from both OpenAI and Anthropic. That mix points to what EAIA is: a reference build you adapt, not a finished product you buy.
The biggest limitation is setup effort. EAIA has no sign-up page and no monthly fee, because there is nothing hosted to sign into by default. You fork the repo, wire up Google OAuth, drop in API keys, and write a configuration file that describes who you are and how you like your email handled. If you want it running around the clock, you also deploy it on the LangGraph Platform, which needs a paid LangSmith plan.
Getting Started
- Fork and clone the
executive-ai-assistantrepository, create a Python virtual environment, and runpip install -e .. - Add your OpenAI and Anthropic API keys, then enable the Gmail API in Google Cloud and run
python scripts/setup_gmail.pyto complete the OAuth flow. - Fill in
eaia/main/config.yamlwith your name, timezone, scheduling preferences, and rules for which emails to ignore, flag, or answer. - Run the agent locally with
langgraph dev, then trigger an ingest job to pull recent emails and review the drafts it produces.
Product Information
A quick look at LangChain's pricing, supported platforms, and performance.
Best for
The users, tasks, and scenarios where this tool fits best.
Users
- Developers and technical founders who already juggle a heavy inbox and are comfortable running open-source tools
- Small teams that want email automation on their own infrastructure
- Product builders studying agent design
Tasks
- Triage
- AI email drafting
- Meeting scheduling
Scenarios
- Clearing a backlog after a few days away
- Keeping a busy work inbox in order
- Prototyping an internal assistant
Key features
Gmail AI triage that follows your rules
EAIA sorts every new message into one of three buckets: skip it, tell you about it, or draft a reply. The buckets aren't hard-coded. You describe them in plain language inside the config file, so "ignore newsletters" or "flag anything from my accountant" becomes part of the agent's behavior.
Draft replies in your own voice
When an email needs an answer, EAIA writes a response and saves it as a draft rather than sending it. The rewrite step uses your tone preferences, so the text sounds like you and not like a generic assistant. You still open the draft and hit send.
Calendar-aware scheduling
Ask EAIA to set up a meeting and it looks at your calendar, finds genuinely free slots, and suggests times in its reply. The scheduling logic sits in its own module, so you can point it at a different calendar provider without touching the rest of the agent.
Human-in-the-loop inbox
Every draft lands in Agent Inbox, a companion interface where you review, edit, and approve the agent's work. This approval step is the reason EAIA is safe to leave running: nothing reaches a real recipient until a human clears it.
Memory and reflection
EAIA keeps notes about the people and context it encounters. A reflection step updates that memory as it learns, so replies get more accurate over time instead of resetting with each message.
Runs on LangGraph Platform
For always-on use, EAIA deploys to the LangGraph Platform with a cron job that checks for new mail on a schedule. Deploying there needs a paid LangSmith account, while local runs stay free.
Pros and cons
Pros
- Open-source under the MIT license, so you can read every part of the logic and change it.
- No per-seat email assistant fee; you pay only for the model API calls you make.
- The human approval step in Agent Inbox keeps a person in control of what actually gets sent.
- Config-driven triage rules let you shape behavior without rewriting code.
- Runs on your own infrastructure, which suits teams with strict data rules.
Cons
- Setup is a developer task: OAuth, API keys, and a config file stand between you and a first draft.
- The hosted, always-on path requires a paid LangSmith plan, so "free" mostly means running it yourself.
- It targets Gmail first, so other email providers need extra work.
- You supply and pay for the underlying OpenAI and Anthropic models separately.
Frequently asked questions
Yes. The code is open-source under the MIT license and costs nothing to download or run. Your real costs are the OpenAI and Anthropic API calls plus the LangSmith plan if you deploy it on the LangGraph Platform.
Related content
Explore related tools, skills, and articles for LangChain.
LangChain Alternatives
Google EmbeddingGemma 2
Google DeepMind · Writing · Image · CodingGoogle EmbeddingGemma 2 is a 740M-parameter open model that turns text, code, images, audio, and video frames into vectors in one 768-dimension space, running on phones and laptops.
MePersona
MePersona, Inc. · Writing · Productivity · MarketingMePersona is an AI unified inbox for creators. It learns your voice, drafts replies to fan messages from 12 platforms, and sends only what you approve.
Fluence
Yay Software (Yannick Huchard) · Writing · Voice & Language · ProductivityFluence is an Android app that reads PDFs, EPUBs, Word files and web articles aloud, follows each sentence as it speaks, and answers questions about the text on the phone itself.