Era AI

AI settings: models, budgets, meetings and voice

Choose models, daily budgets, meeting notes and voice limits.

AI settings determine how Era works, which sources it uses and spending limits. Web and desktop: Settings → AI. Subscription/key setup is explained under Sources.

Before you start

AI management and connecting your provider require different access. Inspect the shared setting scope.

Working preferences — /settings/ai

OptionEffect
Model Era usesSource/model picker; default for later answers, separate from saved agent/workflow choices.
Tool calls per answer1–40; bounds operation steps in one answer.
Normally confirmed actionsUnattended-work preference; action, role and approval safeguards still apply.
Key rotation reminder7–730 days; weekly reminders for old keys after provider checks.

Open Agents / Event agents / Runs links open /agents rather than another agent list here.

Providers — /settings/ai/providers

Add a provider, rotate keys, set custom endpoints, choose personal use or disconnect. Model per kind of work selects a main and fallback model. Empty rows use Auto; fallback is tried once when the first model refuses. What each model costs shows catalogue prices and who pays here. Prices are not quotes; some catalogue prices can be estimates.

Era is not a model provider and keeps no key of its own: every call goes straight to the provider on a connected key (your own, one shared with you, or the workspace's). What each call cost is on the usage page; caps are under Budgets.

Budgets — /settings/ai/budgets

Add budget: whole workspace or one agent, daily USD, monthly USD and alert percentage (1–100, default 80). Empty caps mean no limit for that period. Workspace caps bind all work; agent caps narrow further and the tighter limit blocks first. Edit caps/alerts in rows; removing a budget removes that scope's cap. The last-30-days table shows source/model, runs, input/output/cache tokens, tool calls and USD cost.

Meetings — /settings/meetings

My calendars: connect an account, toggle it or disconnect. Era reads the next seven days. Each connection has Era joins / How / Notes go to preferences.

Notetaker: name in the call, notes language, monthly minutes, longest meeting, brief lead time and recording notice. The owner placeholder becomes your name; blank uses Era's notice.

Service: default Recall.ai or supported own-account choice; who pays, API key, webhook address and signing secret. Recordings: retention days (0 = keep) and who sees new meetings besides the owner. Transcripts and notes remain after audio removal. Remembered voices reuses corrected speaker names.

Voice — /settings/voice

Monthly call-minute cap and daily calls Era may place. At the limit callers receive a stopping message. Finished calls are checked every five minutes. For instant delivery on your own ElevenLabs account, register the displayed post-call webhook URL and secret with the provider.

MCP — /settings/integrations/mcp

Connect external tool servers. Do not confuse MCP settings with desktop-local MCP connections.

Who changes these?

AI management uses ai.manage; budgets and service options check their respective module-management permissions. Connecting your own account/key differs from changing shared organization settings. Ask Era “Explain this agent's daily spending limit and chosen model.” See usage.

AI settings: models, budgets, meetings and voice
AI settings: models, budgets, meetings and voice

See the model catalogue for names/capabilities and computer rules for site, step and time limits.

© 2026 Era. Era documentation. Something missing or wrong? Write to us: [email protected]