Native tools
Built-in integration with Cloud, Domains, Mail, and Hosting. The agent never has to be told how to connect - the credentials and schema are automatically provided.
Agent
The agent that builds with you.
A single intelligent agent that natively understands your workflows, queries your databases, reads your files, and orchestrates your deployments - transparently and securely.
34 toolsBuilt into the platform
All mediaText, image, audio, video
1 contextEverything in one place
What you getAn agent that
An agent that
actually does the work.
It does not just write code for you to paste. It reads your live databases, deploys your applications, and sends your emails.
Web and Search
It can search the live web, read pages, and synthesize answers from multiple sources in real time.
File handling
Upload images, PDFs, audio, or video. The agent understands multimodal input and can generate or edit code files directly.
Safe by default
Destructive actions require explicit approval. The agent will always ask you before dropping a table, deleting a file, or sending an email.
ModelsThe right model,
The right model,
for every task.
Access the most capable AI models directly in your workflows. The agent intelligently routes your queries to the best engine for the job.
ModelProvider
Volume tasksGPT-6 LunaOpenAI
Standard tasksGemini 3.8 FlashGoogle
High-endClaude Sonnet 5Anthropic
Long reasoningGPT-6 AstraOpenAI
Long reasoningClaude Opus 5Anthropic
Long contextKimi K3Moonshot
All models are integrated natively. Failed requests are always refunded in full.See full pricing
Questions
Agent, in detail.
Does the agent cost extra?
No. The agent, its tools, and its multimodal capabilities are included on every plan, including the free one. You simply pay for the underlying usage credits.
What can the agent access?
Only what you explicitly permit. It can read your managed databases and connected storages, but it will always ask for approval before executing a write or delete operation.
Can I change the model?
Yes, the agent can transparently route requests to different models depending on the complexity of the task and the required capabilities.