How Agently Works

Chatting with Jarvis

Chat is the fastest way to work with Jarvis. You describe what you need in plain language, and the agent uses its tools, your Brain, and your connected apps to get it done, right there in the Command Center. For quick, exploratory, or one-off work, chat is where you'll live. (For durable, repeatable work, turn it into a Mission instead.)

Writing Effective Prompts

Jarvis works best when you're clear about the outcome. Three principles:

Describe the outcome, not the step.

  • Instead of "help me with sales"
  • Try "Research Acme Corp and draft a cold outreach email to their VP of Engineering, focused on how we help dev teams ship faster."

Give it context. Reference specific items with @-mentions, name the details that matter (timeline, audience, tone, constraints), and attach files when the agent needs to see something.

Let it run, then steer. Jarvis chains many steps on its own, so you can hand it a whole task ("research, draft, and schedule the follow-up") and course-correct from its output rather than micromanaging each step.

@-Mentions: Pull in Exact Context

Type @ in the composer to reference something from your workspace directly:

  • Pages, point Jarvis at a specific document to read and use
  • Brain items, reference an exact piece of company knowledge
  • Tasks, bring in a task from Spaces for context
  • People, reference a teammate

When you @-mention something, Jarvis gets that item's content directly, faster and more precise than describing it and hoping the agent finds it.

Example: "Draft a follow-up to the client based on @Meeting Notes – Jan 15 and reference the pricing in @Product Pricing 2026."

/-Skills: Invoke a Playbook

Type / to run a Skill by name, a saved playbook that tells Jarvis how to do a specific job your way (draft an X post in your voice, turn research into a cited digest, and so on). You can also just describe the task; if an installed skill fits, Jarvis applies it on its own.

Attaching Files

Click the attachment icon or drag and drop to bring files into the conversation:

  • Documents (PDF, DOCX, TXT), Jarvis extracts and reads the text
  • Images (PNG, JPG), Jarvis can view and analyze them

Attachments are ideal for one-off inputs that aren't in your Brain yet, a contract you just received, a screenshot, a report someone emailed.

Watching Jarvis Work

The chat is transparent. As Jarvis works, you see its progress live:

  • A tool timeline, each action (searching the Brain, sending an email, creating a task, searching the web, reading your calendar, drafting a page) shows as a status row.
  • Delegation cards, when Jarvis runs a Skill or hands a heavy sub-task to a specialist sub-agent, you see it.
  • Citations, facts pulled from your Brain show their sources, so you can trust and verify.

Managing Conversations

  • History, conversations save automatically; return to any thread from the sidebar to pick up where you left off.
  • Keep related work together, context earlier in a thread makes later answers better, so keep a task in one conversation.
  • Start fresh when it drifts, if a thread has gone off track, start a new one rather than fighting to redirect it.

Tips

  1. Reference your Brain, the more knowledge you've added, the sharper the answers.
  2. Be specific about format, want a table, bullet points, a formal email? Say so.
  3. Iterate, treat Jarvis like a collaborator: review, give feedback, ask for a revision.
  4. Use @-mentions liberally, direct references beat hoping it finds the right context.
  5. Promote repeat work to a Mission, if you'll run it again, save it as a Mission or a Skill.
← SkillsBrain →