AI assistant for email management

Email is where good intentions go to die. The average professional spends hours each week reading, triaging, and replying — and the longer a thread runs, the harder it is to remember who promised what. MiyoMind turns inbox management into a short conversation: you tell Miyo what you need in plain language, and it reads the thread, drafts the response, and reminds you to follow up. You stay in control of what actually gets sent.

How does an AI email assistant actually help?

MiyoMind connects to your existing email through Gmail or Microsoft Outlook over secure OAuth, so you never paste a password. Once connected, you talk to Miyo the same way you'd ask a sharp assistant for help. It handles the three things that eat the most time: understanding what a thread is about, writing a reply, and making sure nothing slips.

• Summarise long threads: "What's the latest on the Henderson contract email chain — who's waiting on what?"
• Draft replies in seconds: "Reply to the recruiter and politely decline, but ask to stay in touch for senior roles."
• Catch the to-dos buried in your inbox: "Pull out every action item people emailed me this week."
• Never forget to follow up: "Remind me Thursday morning to chase the invoice if the client hasn't replied."
• Triage at a glance: "Summarise my unread emails from today in three bullets each."

Because Miyo also has long-term memory, it learns the context that matters — your tone, the projects you care about, the people you work with most — so its drafts get more useful over time rather than generic on every request.

What can I actually ask it to do? (Example prompts)

Here are real messages you'd type into WhatsApp, Telegram, Discord, or the MiyoMind web dashboard — and what Miyo does in response.

You: Summarise the thread with Priya about the Q3 launch.
Miyo: Priya confirmed the launch date is July 14. She's waiting on
  your final copy by July 2, and asked whether legal signed off. No
  reply from you yet on the legal question.

You: Draft a reply: confirm July 14 works, copy by July 1, legal
  approved last week.
Miyo: [drafts a ready-to-send reply for you to review and send]

You: Remind me July 1 at 9am to send Priya the launch copy.
Miyo: Done — recurring? Or one-off?

You: One-off is fine.
Miyo: Set. I'll ping you here on July 1 at 9am.

Notice the division of labour: Miyo reads, summarises, and drafts — then hands the finished reply back to you. You send it from your own email client (or copy it across). The reminder fires in whatever chat app you're using, so the follow-up actually happens.

Drafting vs sending: where's the line?

This is the honest part, and it matters. MiyoMind's job in the email workflow is to read your messages and produce drafts and summaries. It is not an autopilot that fires emails into the world without you. Here's how the responsibilities split:

Keeping the send button with you is deliberate. It means a drafting mistake is never an outbound mistake, and you stay accountable for every message that leaves your name. It also keeps you firmly inside your email provider's own terms — Miyo helps you write faster without acting as you.

Is it safe to connect my email?

Email is sensitive, so the security model is built around it. Connections use OAuth — the same delegated-access standard your phone uses to link apps — so MiyoMind never sees or stores your email password. The access tokens that authorise the connection are encrypted at rest with AES-256-GCM.

• Every paid user runs inside their own dedicated, sandboxed Docker container with an isolated workspace, no public internet egress, a read-only root filesystem, and dropped Linux capabilities.
• Integration tokens and your long-term memories are encrypted at rest with AES-256-GCM.
• A 10-layer prompt-injection defence plus output scrubbing runs on every message — so a malicious email can't trick Miyo into leaking data or taking rogue actions.
• You can disconnect Gmail or Outlook at any time from the dashboard.

MiyoMind runs the open-source OpenClaw agent runtime, a model router called Hermes, and our own orchestration, memory, billing and safety code — drawing on frontier models from OpenAI, Anthropic, Google, xAI and Alibaba. The intelligence comes from the best available models; the privacy and routing are ours.

A realistic morning with Miyo

Say it's Monday at 8:40am and you're on your phone. You open WhatsApp and message Miyo: "Summarise my unread from the weekend, three bullets each, flag anything urgent." Thirty seconds later you've triaged twelve emails without opening one. You reply: "Draft a no to the speaking request, yes to the partner call, ask for Tuesday or Thursday." Two drafts come back; you tweak one word and send them from Gmail. Last message: "Remind me Wednesday 2pm to follow up with the partner if I haven't heard back." By 8:45 your inbox is handled and the loose ends are on a timer. The same flow works from your laptop in the web dashboard, or from Telegram and Discord.

What does it cost?

MiyoMind is credit-based, and credits meter the actual model and tool usage behind each request — so a quick summary costs far less than a long research task. There's a free tier to try the email workflow before you commit.

Top-up credit packs are also available (600 for $3, 2,000 for $10, 5,000 for $25, 10,000 for $50), and 1 credit is worth roughly $0.005 of value. The free tier runs on a shared direct-agent path; paid plans give you your own sandboxed container, which is what you want once email and integrations are involved.