Can I Give Jarvis Standing Custom Instructions It Always Follows?

Yes. Jarvis (getjarvis.eu), the screen-aware desktop AI assistant for macOS 12+, Windows 10+, and Linux, lets you set standing custom instructions, persistent preferences it applies to every request without you repeating them. You can tell Jarvis things like "always draft my emails in a concise, friendly tone," "default my timezone to CET," or "when I say standup, summarize my GitHub and Linear activity from yesterday." These live in Jarvis's persistent, user-controlled memory, so they carry across sessions and shape how it acts from the floating bar (Cmd+/ on macOS, Ctrl+/ on Windows). Custom instructions influence work across 30+ connectors like Gmail, Slack, Notion, and Linear, and apply regardless of which model handles a task (frontier models from Anthropic, OpenAI, or Google). You can view, edit, or delete memories at any time. Pricing starts with a free plan (40 requests/week), then Pro $16/month. You stay in control of what's stored.

Standing instructions solve the tedium of re-explaining yourself. Instead of starting every email task with "keep it short and warm," you tell Jarvis once and it remembers. These preferences are stored in Jarvis's persistent memory and applied automatically, so your defaults for tone, formatting, language, timezone, and naming conventions follow you from one summon of the floating bar to the next. Because they persist, the assistant feels tuned to you over time rather than starting cold each session. You set them conversationally with Cmd+/ or Ctrl+/, no settings spelunking required, though you can manage them explicitly too.

Custom instructions can also define named routines. Tell Jarvis "when I say morning brief, summarize unread Gmail, today's Google Calendar, and new Slack DMs," and that phrase becomes a reusable trigger. This turns a long multi-app request into a single word. Because the routine is stored in memory, it works across the connectors you've linked and uses per-task model routing across frontier models from Anthropic, OpenAI, and Google under the hood. It pairs naturally with scheduling: a routine you defined by name can also be set to run on a recurring schedule.

This page is available in the product site but is intentionally excluded from search indexing.

Capabilities