One provision call and your agent operates on its own — real email inbox, phone number, web search, file storage, page hosting, 2FA codes, cron scheduling, encrypted vault, and a synthetic persona. All on domains and numbers reserved exclusively for your project. No human in the loop.
Add as remote MCP endpoint in Claude, Cursor, or any MCP client
ext install keyid.keyid-agent-inboxOr search "KeyID Agent Inbox" in the VS Code Marketplace
npm install @keyid/sdkThen: const agent = new KeyID(); await agent.provision();
pip install keyidThen: agent = KeyID(); identity = agent.provision()
POST https://keyid.ai/api/v1/provisionRequires a project key — use from any language or runtime
How it works
Each project gets domains and phone numbers reserved exclusively to it. Every agent inside a project gets its own identity and inbox, but the sending reputation belongs to your project alone — nobody else's agents can ever touch it. Domains are warmed before they carry traffic.
Provisioning requires a project key. The open shared pool is closed: it attracted bulk account-farming rather than real deployments, so access is now issued deliberately, per project. That keeps deliverability predictable for the fleets actually running on it.
Dedicated infrastructure
Every deployment is dedicated. For teams that need isolated sending reputation — serious outreach, customer-facing fleets, brand-sensitive sending — we reserve domains and phone numbers exclusively for your project. Nobody else's agents can ever touch them.
Reserved domains & numbers
Your domains and SMS numbers are fenced off from the shared pool. Your fleet provisions with a private project key; everyone without it stays on shared infrastructure.
Your sending and verification identity is yours alone.
Managed warmup
We configure mail records, verify the domain with the email provider, and run a gradual warmup schedule with real engagement before your campaigns start.
Domains arrive campaign-ready, not cold.
Fleet-ready sending
Project fleets skip newcomer throttles and template-similarity blocks that protect the shared pool, and get custom per-agent rate limits sized to your campaign.
One config line in the SDK: your project key.
Pick a plan and your project key is issued immediately, with a warmed domain already reserved to it. Need an invoice, a PO, or a DPA? Talk to us instead.
Pricing
Every plan includes unlimited agents on your own domains — email, phone, web search, file storage, page hosting, 2FA, cron, and encrypted vault. You are buying isolated sending reputation, not a seat count.
Starter · soon
1 reserved domain · unlimited agents · 5,000 emails/month
Full autonomy stack included.
Growth · most popular · soon
3 reserved domains · unlimited agents · 25,000 emails/month · 1 phone number for SMS and OTP
Priority warmup on new domains.
Scale · soon
10 reserved domains · unlimited agents · 100,000 emails/month · 3 phone numbers
Direct support.
New domains arrive warming. Your reserved domain is registered fresh for you and completes a 14-day warmup that gradually builds sending reputation. Mail to your own company's servers works immediately; large external sends improve steadily over the first two weeks. We run the warmup for you — nothing to configure.
Need more domains, numbers, or volume than a plan includes? Email hello@keyid.ai and we will size it for you. Prices exclude VAT, applied automatically at checkout.
Already a customer? Manage or cancel your subscription — enter your billing email and we'll send you a sign-in link. No account or password needed.
Use cases at scale
The common pattern is not one blocked registration. It is repeated account creation, outbound email, replies, support, and recovery across a fleet where each agent needs its own inbox identity.
Outbound fleets
Give campaign agents, account managers, or follow-up workers their own inboxes so sending, reply handling, and recovery do not collapse into one shared mailbox.
Each agent keeps its own send, receive, and recovery identity.
Registration waves
When agents need accounts on marketplaces, tools, forums, or partner systems, the same identity layer can catch email links, SMS codes, and login checks without a human loop.
One path covers inboxes, phone access, links, codes, and saved state.
Support and reply teams
Customer-facing fleets need separate identities for replies, follow-up, and recovery, otherwise every message thread piles into a human-owned shared mailbox.
Replies stay attached to the right agent identity.
Agents normally need human sign-ups for each of these services. KeyID bundles them behind a single identity with no API keys and no registration.
Email sending
Replaces SendGrid, Mailgun, Postmark, Amazon SES
Email accounts
Replaces Gmail, Google Workspace, AgentMail, Outlook
SMS & phone
Replaces Twilio, Vonage, Plivo
Web search
Replaces Serper, Tavily, Brave Search, SerpAPI, Google Custom Search
File storage
Replaces AWS S3, Google Cloud Storage, UploadThing
2FA & TOTP
Replaces Google Authenticator, Authy, 1Password TOTP setup
Verification
Replaces manual email checking, link clicking, code copying
Scheduling
Replaces cron-job.org, AWS EventBridge, Inngest, Trigger.dev
Static hosting
Replaces Vercel, Netlify, GitHub Pages, CloudFlare Pages
Answers teams ask before they move one agent or a whole fleet onto real inboxes, SMS, and website verification work.
Yes. Pick a plan and your project key is issued immediately, with a warmed domain already reserved to it — no waiting on us. The open shared pool stays closed: access is tied to a paid project, which is what keeps other people's agents off your sending reputation. If you need an invoice, a PO, or a DPA, contact us and we will handle onboarding directly.
Fleets. A project can run hundreds or thousands of agents that need separate inboxes for registration, sending, reply, and recovery — all on domains reserved to that project.
Yes. KeyID is built for the real website path: inboxes, phone and SMS, magic links, email codes, and TOTP so agents can keep moving through third-party registration and recovery flows.
Those paths either assume human-owned accounts, paid mailbox seats, API-key setup, or pricing that gets painful once you need a real fleet. KeyID is aimed at agent-native infrastructure instead. The full breakdown is on the compare page.
No. The goal is zero human setup in the critical path. Agents can get real communication identity without first making your team provision and manage a human mailbox stack.
See the For agents section below. The setup and runtime references live there and on the linked docs pages.
Yes. The same infrastructure that handles registration and verification also supports real inbound and outbound communication, which is why it works for support agents, outreach fleets, and business-facing automation too.
Implementation details live here. Expand for hosted MCP, local stdio, and runtime guides for Claude, Cursor, Lovable, Replit, OpenClaw, and Playwright.
Remote
https://keyid.ai/mcpLocal
npx -y @keyid/agent-kitClaude + Cursor
Connect the hosted endpoint or the local package, then give the agent inbox, SMS, and website verification access.
Lovable + Replit
Use SDK or API calls in app code, then keep website registration and verification loops on real inboxes and real phone/SMS channels.
OpenClaw
Connect KeyID as a remote MCP skill so OpenClaw agents can provision inboxes, receive verification codes, and handle OTP flows across a multi-agent fleet.
Playwright
Let Playwright drive the site while KeyID handles inboxes, SMS, codes, and persistent website state for long third-party flows.
Private infrastructure. Full autonomy. No human in the loop.
Email, phone, search, storage, hosting, 2FA, scheduling, and vault — inside a project you control. No human creates it. No human manages it. That's autonomy.
Contact us
We use this only to follow up about private infrastructure, larger deployments, or custom needs.