AI Gateway
One endpoint for Claude Code, Cursor, VS Code and any OpenAI-compatible tool. List your AIs in order; when one is rate-limited or down, the next one answers. Spending limits per provider.
AI Gateway · Automation · AI Agents
Your AI, always on.
Hosting for the AI tools you already use. Claude Code and Cursor keep working when a model is rate-limited, your automations run around the clock, and you never maintain a server.
At a glance
- AI Gateway
- from $2.40 / month
- Automation
- from $7.20 / month
- Works with
- Claude Code · Cursor · VS Code · any OpenAI-compatible tool
- AI providers
- OpenRouter · Anthropic · OpenAI · Gemini · DeepSeek · Groq
- Your keys
- You pay the AI providers directly
Products
Three ways to put your AI to work.
Each one runs on our servers, 24/7, with your own AI keys.
Automation
Your own Activepieces with hundreds of apps: Gmail, Google Sheets, WhatsApp, Slack, GitHub, social media and CRMs. Add AI steps, or ask your AI to build the flow for you.
AI agents (soon)
Hermes and OpenClaw: a personal AI agent you chat with on Telegram, in your editor and in your terminal. It browses, runs tools and remembers. Join the waitlist.
AI Gateway
Rate-limited? Your next AI answers.
Start on free models, keep a cheap one as backup and a strong one for hard tasks. The gateway picks the next AI on your list the moment one fails, so your session keeps going.
- 01
Add your AIs
Paste your keys and put the models in order, for example OpenRouter free, then DeepSeek, then Claude.
- 02
Set limits
Cap spending per provider for every 5 hours and every week. Over the limit, the gateway moves on to the next AI.
- 03
Paste one URL
The dashboard shows the exact settings for Claude Code, Cursor, VS Code Copilot and OpenAI SDKs. Copy, paste, done.
Claude Code ready
Native Anthropic endpoint for Claude Code, plus a model per task: background work on a free model, thinking and long context on your strongest one.
OpenAI-compatible
Cursor, VS Code Copilot (custom endpoint), Hermes, LangChain and any OpenAI SDK use the same gateway with one token.
Your keys stay yours
Keys are stored encrypted and only used to call the providers you chose. Claude Pro or Max subscriptions never go through the gateway: keep using them directly and let the gateway be your backup.
Recipes
Copy a prompt. Your AI builds the flow.
Paste a recipe into Claude, the AI in your editor or your agent while it is connected to your Automation. It creates the flow; you connect your apps and switch it on.
Marketing
Blog post to social media
Every new post on your blog becomes a short post on X and LinkedIn.
See prompt
In my Automation, create a flow: when my blog's RSS feed has a new item, use an AI step to write a short, friendly post (under 250 characters, 2 hashtags) with the link, then publish it to X and to my LinkedIn profile. Ask me for any app logins you need. Show me the flow and test it once before turning it on.
Website leads to a sheet + welcome email
New form leads are logged and get a personal reply within a minute.
See prompt
In my Automation, create a flow with a webhook trigger for my website form (name, email, message). Add each lead as a row in a Google Sheet called Leads, then send the lead a short welcome email from my Gmail that an AI step personalises from their message. Ask me for any app logins you need. Show me the flow and test it once before turning it on.
Content calendar autoposting
Write posts in a Google Sheet; they go out at the time you set.
See prompt
In my Automation, create a flow that runs every 15 minutes, reads my Google Sheet Content Calendar (columns: date-time, platform, text, status), publishes rows that are due and not posted yet to the platform named in the row (X or LinkedIn), and sets their status to Posted. Ask me for any app logins you need. Show me the flow and test it once before turning it on.
Development
GitHub issue triage to Telegram
New issues are summarised and labelled, and you get a ping.
See prompt
In my Automation, create a flow: when a new issue is opened in my GitHub repo OWNER/REPO, use an AI step to summarise it in 2 sentences and suggest one label (bug, feature or question), add that label to the issue, and send the summary with the link to me on Telegram. Ask me for any app logins you need. Show me the flow and test it once before turning it on.
AI first look at pull requests
Each new PR gets a short AI review comment.
See prompt
In my Automation, create a flow: when a pull request is opened in my GitHub repo OWNER/REPO, fetch its title, description and changed files, use an AI step to write a short review (risks, missing tests, questions), and post it as a comment on the pull request. Ask me for any app logins you need. Show me the flow and test it once before turning it on.
Error alerts to Telegram
Your server or uptime monitor calls a webhook; you get a clear alert.
See prompt
In my Automation, create a flow with a webhook trigger that my server or uptime monitor calls on errors. Send me a Telegram message with the service name, the error and the time, and log every alert as a row in a Google Sheet called Incidents. Ask me for any app logins you need. Show me the flow and test it once before turning it on.
Operations
Invoices from email to a sheet
Invoice emails are read and the numbers land in your sheet.
See prompt
In my Automation, create a flow: when a new Gmail email arrives with an invoice or receipt, use an AI step to extract vendor, date, amount and currency, and add them as a row in a Google Sheet called Expenses. Ask me for any app logins you need. Show me the flow and test it once before turning it on.
Daily inbox digest
One morning message instead of checking email all day.
See prompt
In my Automation, create a flow that runs every day at 07:00: collect my unread Gmail emails from the last 24 hours, use an AI step to summarise them in a short list ordered by importance, and send the list to me on Telegram. Ask me for any app logins you need. Show me the flow and test it once before turning it on.
Pricing
Pay for hosting. Your AI usage stays with your provider.
AI Gateway
Backup AIs for Claude Code, Cursor and more.
from$2.40/ month
$3.00 monthly$2.85 / mo on 3 months$2.70 / mo on 6 months$2.40 / mo on 12 months
- One endpoint, OpenAI and Anthropic formats
- Automatic fallback when a model is rate-limited or down
- Spending limits per provider (5 hours / week)
- A model per Claude Code task
- Always on, nothing to install
Automation
Hosted Activepieces for any workflow.
from$7.20/ month
$9.00 monthly$8.55 / mo on 3 months$8.10 / mo on 6 months$7.20 / mo on 12 months
- Your own Activepieces instance, hundreds of apps
- Marketing, development and operations flows
- Build flows by asking Claude, your editor or your agent
- Always on, no server to run
Hermes agent
Coming soonA personal AI agent on Telegram, in your IDE and in your terminal.
from$5.04/ month
$6.30 monthly$5.99 / mo on 3 months$5.67 / mo on 6 months$5.04 / mo on 12 months
- Chat on Telegram, paired in one step
- Multi-agent: up to 5 sub-agents working in parallel
- Use it from VS Code, Antigravity or Cursor over MCP
- Webhooks: GitHub or a form can start a task
- 5 GB storage
Prices are for hosting. AI usage is billed to you by the AI provider you choose. Pay monthly or save with a 3, 6 or 12-month plan. USDT or PayPal.
FAQ
Questions, answered.
What does the AI Gateway do?
It gives your AI tools one address. Behind it is your list of AI models. When the first one is rate-limited, out of budget or down, the request goes to the next one automatically, so Claude Code or Cursor keeps working instead of stopping.
Can I use my Claude Pro or Max subscription through it?
No, and on purpose: Anthropic's terms don't allow subscriptions to be shared through a service like this. Keep using your subscription in Claude Code directly. When you hit its usage limit, switch that terminal to the gateway (the dashboard shows the two lines to paste) and continue the same conversation on your own API keys.
Do I need paid AI keys?
Not to start. OpenRouter's free models work as the first step of the chain. They are rate-limited, which is exactly when the gateway hands over to your next AI. For daily use, add a cheap paid model such as DeepSeek as backup.
Who pays for AI usage?
You, directly to each AI provider, with your own keys. Our price covers the hosting only. Spending limits per provider help you keep that bill predictable.
Do I need to code to use Automation?
No. Describe the flow you want to your AI, or copy one of the recipes above, and it builds the flow in your Automation. You can also build flows by hand in the visual editor. You connect each app (Gmail, Sheets, Telegram...) once, inside Automation.
Which AI can build automation flows for me?
Claude (claude.ai) as a connector, the AI in VS Code, Antigravity, Cursor or Claude Code over MCP (the dashboard shows the config to paste), or a hosted agent. The flows then run on our server on their own, without the AI.
Is Hermes multi-agent?
Yes. For a bigger job Hermes can split the work across up to 5 sub-agents that run in parallel, each with its own context, and then combine their results. Each sub-agent uses its own AI requests, so costs grow with them, and research quality depends on the model you pick.
When will Hermes and OpenClaw be available?
They're in testing now. Join the waitlist and we'll email you once, at launch.
How do I pay?
Monthly, or 3, 6 or 12 months up front at a discount. USDT (BEP20 or TRC20) or PayPal. A service starts as soon as its invoice is paid, and you can stop at any time by not renewing.
Give your AI a home tonight.
Always-on AI hosting from $2.40 a month. Your keys, your models.