Skip to main content

Frequently Asked Questions

General​

What is FlyMy.AI?​

FlyMy.AI is a cloud platform for AI agents - every layer as pure infrastructure:

  • Agents - describe a worker in plain language (or code); it plans, calls tools, and returns results. Freeze it into a versioned instruction and call it as an API, run it on a schedule, or embed it in your product.
  • Models API - any hosted model (image, video, LLMs, audio) through one unified endpoint. No GPUs to manage.
  • MCP integrations - 1,100+ verified tools (240 native + 900+ via Composio): Slack, Notion, GitHub, Gmail, CRMs, web search and more - callable by agents or directly.
  • Serverless - deploy your own models as auto-scaling endpoints, and scale agent runs without managing workers.

Who is it for?​

Developers shipping agent-powered products, teams automating recurring work, and non-coders who build agents in the web app in plain language. If you resell to your own customers, embedded deployments isolate every customer's connections automatically.

Where do I start?​

Models​

What models are available?​

The catalog moves fast - image (SDXL Turbo, Flux family, Nano Banana, GPT Image), video (Veo, Seedance, Sora), LLMs, speech and more. Browse the live list at app.flymy.ai or query it programmatically (flymyai_models / recommend_model on the MCP gateway). Don't hardcode model lists from docs - ask the API.

Can I deploy my own model?​

Yes - private serverless deployments run your own models as auto-scaling endpoints. Contact us via app.flymy.ai to enable it for your account.

Billing​

How does pricing work?​

Pay per use. No seats, no minimums, no subscription tiers. Agent runs are billed by actual LLM usage and tool calls (the API reports the exact cost of any run); model calls are billed per prediction. See pricing.

Are there rate limits?​

Rate limits apply per account to keep the platform stable. If you hit them, they can be raised - reach out.

Data & security​

Is my data private?​

Your inputs, outputs, and run history are stored in your account so your agents and generation history work - they are yours and are not used to train models. Connection credentials (OAuth tokens, API keys) are encrypted at rest, never exposed to the model, and only decrypted inside an isolated sandbox at execution time.

How are MCP integrations secured?​

Every integration in the catalog is reviewed by the FlyMyAI team - no arbitrary community servers touching your data. Per-user isolation is enforced: one customer's connections can never appear in another customer's run. See Why FlyMyAI.

Agents​

Do agents keep running when I close my laptop?​

Yes. Runs execute in the cloud; scheduled (frozen) agents fire on their cron regardless of your machine.

What's the difference between running and freezing an agent?​

Run = interactive, exploratory. Freeze = the successful run becomes a fixed, versioned instruction you can re-run with fresh variables, schedule, or publish as an API / embed in your product. Recurring work should be frozen - it is faster, cheaper, and deterministic. See Runs & Freeze.

Can my customers use my agent with their own accounts?​

Yes - that is an embedded deployment: one frozen agent, every customer connects their own Slack/Gmail/CRM, fully isolated by external_user_id. See Embed an Agent in Your Product.

Support​

Where do I get help?​