Skip to main content

Your first day building agents

The editor journey — create an agent, give it instructions, and watch it do real work on a task.

2 min read

This journey is for the person who turns "the team keeps asking the same questions" into an agent that answers them. In fifteen minutes you create an agent, shape how it behaves, and watch it do real work on a task — the loop every later agent refines.

You need the Editor role or higher (the Agents section is hidden from members) on a workspace where chat already answers — that is the quickstart.

  1. Create the agent

    To start an agent teammates can put to work, open Agents in the sidebar and click Create agent. Name it for the job, not the technology — "Support Triage" beats "GPT Helper" — because the name is what teammates assign tasks to later.

  2. Shape its identity

    The editor opens on the General tab: the display name teammates see and a one-line description. The setting that matters on day one is the visibility — it decides who in the organisation may put the agent to work.

  3. Write the instructions

    Open Instructions — the knob that matters most. Write one paragraph as if briefing a new colleague: the voice to answer in, the domain it owns, and the cases it should refuse. Concrete beats complete — you will refine after seeing real replies. Click Save; the agent is reachable from the next request, with no separate publish step.

  4. Watch it work

    Agents do their work on tasks — chat runs the built-in assistant only. Open a project, add your agent under its Agents tab, then create a task that states the work in one sentence and assign it to the agent. Start the run and watch its timeline; the result comes back for your review, and only you can mark it done.

Where you are now

You have shipped the smallest real agent: instructions and a place in the org's roster. The full model behind what you touched is Agent concepts — instructions, knowledge, tools, and skills. The natural next build is your first agent end to end, which adds knowledge bindings and a real domain; after that, agents with knowledge and delegation between agents take the same loop further.

© 2026 Tale by Ruler GmbH — ISO 27001 & SOC 2 certified.

Tale is MIT licensed — free to use, modify, and distribute.