Operators

One clear capability for every workflow.

Friday separates conversation, tools, voice, media, automation, and memory into legible capabilities, each with its own provider, model, and boundaries.

11Workflow catalog
11Live workflows
ChoiceConfigurable providers

Distinct jobs. One assistant.

Each workflow keeps a clear purpose and composes with the others when a request needs more than one step.

01Live

Assistant chat

The main conversational agent. It answers in Markdown, shows tool activity, handles attachments, and saves local session history.

  • Plain-language requests
  • Visible tool activity
  • Markdown, code blocks, and inline images
  • Local conversation history
Supported providers

Assistant provider · Configured model · Local history

02Live

Agent tools

The action surface Friday uses to read and write files, run commands, search or browse the web, generate images, schedule work, load skills, and call MCP tools.

  • Permission prompts for sensitive actions
  • File, shell, web, browser, image, schedule, skill, subagent, and MCP tools
  • Tool output treated as evidence
  • No account-changing action without authorization
Supported providers

Local tools · MCP servers · Skills

03Live

Voice

Dictate prompts with speech-to-text and have Friday read replies aloud with the voice provider and model you choose.

  • Real-time dictation
  • Audio transcription
  • Read-aloud replies
  • Provider and voice model selection
Supported providers

Speech-to-Text · Text-to-Speech · Microphone permission

04Live

Image generation

Create images from text descriptions in chat or a dedicated image workspace.

  • Generated images appear in conversation
  • Save generated images
  • Provider and model selection
  • Available as a chat tool and settings workspace
Supported providers

Text-to-Image provider · Configured model

05Live

Skills

Reusable packaged workflows that teach Friday how to handle a class of work.

  • Install, import, enable, and disable skills
  • Automatic discovery before answering
  • Declared tool and timeout limits
  • Selection and rejection audit trail
Supported providers

Personal skill library · Enabled skills

06Live

MCP servers

Connect remote HTTP or local stdio Model Context Protocol servers so their tools become available to the assistant.

  • HTTP servers with optional token or key auth
  • Local process servers
  • External tools without changing Friday
  • Tools exposed automatically after connection
Supported providers

Remote MCP · Local stdio MCP

07Live

Channels

Talk to Friday from Telegram and Discord today and receive replies back through the same channel. Slack, WhatsApp, and Email are planned adapters.

  • Bot-based messaging
  • Reliable delivery and reconnection
  • Allowlist, pairing, open, or deny DM policy
  • Separate channel secrets and enabled state
Supported providers

Telegram · Discord · Slack coming soon · WhatsApp coming soon · Email coming soon

08Live

Task scheduler

Create recurring assistant jobs such as summaries, checks, and routine cleanup tasks.

  • Create, update, pause, resume, list, and run schedules
  • Each schedule runs an agent turn
  • Provider and model selection for scheduled work
  • On-demand manual runs
Supported providers

Cron schedule · Scheduled assistant model

09Live

Health checks

Run Friday periodically against a checklist you define and report back when something needs attention.

  • Periodic checklist-driven runs
  • Configurable interval
  • Provider and model selection
  • Proactive status reports
Supported providers

Checklist · Interval · Health provider/model

10Live

Memory and personalization

Durable facts, preferences, identity, and workspace context help Friday respond with useful personal context.

  • Durable facts and preferences
  • User identity and tone context
  • Local workspace for produced files
  • Personalization stays on your device
Supported providers

Local memory · User profile · Workspace

11Live

Video and audio generation

Generate video, music, and sound effects from text with the provider and model you choose. Results are playable inline and saved to your local media library.

  • Text-to-video with your chosen provider and model
  • Music and sound-effect generation
  • Separate provider and model choices per media type
  • Saved to the local media library and playable in chat
Supported providers

Text-to-Video · Music / Audio

Bring your own provider

Every operator is provider-agnostic.

Configure your preferred provider and model in Settings. Credentials stay local; the operator resolves them at runtime.