Question 1

How much does the m8tes API cost?

Accepted Answer

The API is pay as you go with no platform fee. Top up a prepaid balance, then each run draws it down at the model provider's list price.

Question 2

How do I fund API runs?

Accepted Answer

New accounts start with a $0 balance. Add funds before the first run. A Platform model subscription does not fund multi-tenant API usage.

Question 3

Can I embed AI agents in my own product?

Accepted Answer

Yes. The SDK supports multi-tenant AI agents with per-user isolation. Set user_id on runs to scope memory, tool access, and task history to individual end users. Your users get their own AI agent experience inside your app, and webhook events notify your product when a run completes or needs approval. We own the infrastructure layer (runtime, integrations, memory, scheduling, observability) so embedding a reliable autonomous agent isn't a backend project of its own. Our own Platform, where teams put agents on real work, runs on the same API.

Question 4

What apps can AI agents connect to?

Accepted Answer

150+ apps including Gmail, Slack, Notion, Google Ads, HubSpot, Linear, Stripe, Google Sheets, and Search Console. All connections use OAuth with scoped permissions. Token refresh is automatic. You can rotate or revoke your API key in one call.

Question 5

How do AI agent permissions work?

Accepted Answer

A human stays on every run. Three modes: autonomous (runs without asking), approval (asks before every sensitive action), and plan (reviews the full plan before execution). You choose the mode per agent, and risky actions can route to the right owner. Every run produces full execution logs.

Question 6

Do you support zero data retention (ZDR)?

Accepted Answer

Yes, API accounts can enable zero data retention so m8tes never stores conversation content, tool inputs/outputs, or reports, only run metadata (status, timing, token and cost counts, which tools ran). Separately, we have Zero-Data-Retention agreements with specific model providers (not every provider), and we route eligible models through those ZDR paths so those providers don't retain your prompts or outputs. Use a ZDR-backed model plus account-level metadata_only for end-to-end zero retention. On every path, your data is never used to train models.

Ship cloud agents. Skip the infrastructure.

From zero to deployed agent in seconds

From weeks of engineering to seconds

Ship it behind your own UI. Every user gets their own agent.

FAQ