Auxot
Tutorials Docs Pricing Login

Tutorial 01

Say hello to the Admin Agent

Send your first message to the Admin Agent — your built-in teammate whose job is helping you build the agents your business actually needs.

Plus: three prompts that reveal the Admin Agent isn't a chatbot — it's the guide who orients you to Auxot and helps you figure out which agents to build next.

Audience Everyone
Time ~2 min
Prerequisites An Auxot account. An AI model connected to it (ChatGPT, Claude, Grok, or similar). If you're on the Free tier, you'll have connected this yourself during signup (or you'll do it in Tutorial 09). On Team, Business, or Enterprise, whoever set up your account has usually connected one already.
You'll end up with A working chat session — plus three prompts you'll keep coming back to.

Why this matters

Your Auxot account comes with the Admin Agent — a teammate, ready to go.

Every business has work that keeps slipping — the research nobody has time for, the follow-ups that fall through cracks, the roles you’ve been meaning to fill. The Admin Agent’s job isn’t to do that work himself. His job is to help you build the agents that will — one specialized teammate at a time, asking the right questions until each new agent knows exactly what it’s there for.

The agents you build will each get their own context files — plain-English documents you provide, like your sales procedures, financial statements, brand voice, or job manuals. Once a custom agent has its files attached, it stops answering as generic AI and starts answering as if it works at your company. (When you’re ready to add your first context file, head to Tutorial 04: Add your first context file.)

Today, send a message and see a reply come back. Tomorrow, ask the Admin Agent to help you build the first agent your business has been waiting for.

You tell the Admin Agent what work you want offloaded — it asks the right questions, helps you build a custom agent for it, and points you at the right places in Auxot to set things up. Each custom agent you build then handles the work you described, with its own context files and tools.

Quick start

  1. Sign in — open Auxot in your browser and log in.
  2. Open chat — click Chat in the left menu.
  3. Select the Admin Agent — make sure the agent picker at the top reads “Admin Agent.”
  4. Send a message — type hello and press Enter.
  5. Watch the reply come in — words should start appearing within a second or two.

Done? You should see a reply from the Admin Agent appear below your message, typed out word by word.

The agent can do that?

You said hello. These three prompts each test a different kind of work the Admin Agent can take on — and if any come back thin, you’ve just found the first thing worth improving.

1. Ask it to audit your own setup

Give me a prioritized punch list of what's set up in my Auxot account right now and what's missing or risky. Be specific. Name the AI models, agents, users, and anything that looks half-done.

Why it’s non-obvious: Most people treat the Admin Agent like a FAQ bot. It actually knows what’s going on in your account — this prompt turns “hello” into a 30-second review that most teams would pay a consultant to do. If the Admin Agent can’t answer this, its instructions need work (that’s Tutorial 05).

2. Ask it to diagnose before you do

Look at the last 24 hours of activity and the health of anything connected to Auxot. What's the single most concerning pattern — not the loudest, the most concerning? Explain why.

Why it’s non-obvious: Pulls chat, system health, and activity history together in one ask — no flipping between three tabs to piece it together. The “not the loudest, the most concerning” framing is the trick; it forces real prioritization instead of a list-of-everything dump.

3. Ask it to push back on you

I'm about to [invite my first teammate as an admin / connect a second AI model / change an agent's instructions]. Talk me out of it, or tell me the one thing I should do first. Be blunt.

Why it’s non-obvious: Inverts the usage. Most people ask agents to do work for them; a well-instructed agent is a better rubber duck than a junior teammate. If the Admin Agent just agrees with you, its instructions are too agreeable — fix that in Tutorial 05.

Go deeper

What the Admin Agent is for

The Admin Agent is the one agent that ships with every Auxot account. It already knows how your account is set up, so you can ask it to answer questions about your setup, walk you through connecting an AI model, help you figure out what’s happening when something misbehaves, or guide you through building your first custom agent. Think of it as the house concierge: always on, always there, with a useful answer the moment you ask — and quick to point you toward the agent you should build instead when your question is really about getting work done. Custom agents you or your team build later will have their own instructions and their own jobs; the Admin Agent is the starting point — and the one that’s there before you’ve built anything.

Troubleshooting

  • A banner about no AI model being connected — Auxot doesn’t have an AI model (like ChatGPT, Claude, or Grok) connected yet. Fix: if you admin the account yourself, connect one in Settings → Providers (see Tutorial 09: Connect a cloud AI model). If someone else admins it, ask them.
  • Message sends but no reply — usually Auxot can’t reach the AI model. Fix: open System Health in the left menu and look for anything flagged red.
  • Reply arrives all at once instead of word-by-word — the live connection between your browser and Auxot isn’t working. Fix: refresh the page. If that doesn’t help, the fix is on the Auxot server side — ask whoever admins your account to look into it.
  • The Admin Agent doesn’t appear in the agent picker — your account doesn’t have permission to use it. Fix: if you admin the account, open Settings → Agents and check your own access. If someone else admins it, ask them (and see Tutorial 06: Invite your first teammate).

Variations & edge cases

  • Already have multiple agents set up? The agent picker remembers your last selection per session. Pinning a default is on the roadmap.
  • Want to try a specific AI model? The Admin Agent uses whichever model is set as the default on your account. To try a different one, make a test agent tied to that model (see Tutorial 07: Create an agent from scratch).
  • Chat history: Your conversations are saved and only visible to you. Clearing history happens one conversation at a time, not all at once.

Walkthrough

Step 1: Sign in

Open Auxot in your browser and sign in. However you got here — you signed up on your own, or a teammate invited you — you just need a working login.

Tip: If you land on a setup page instead of the login screen, your account hasn’t finished its first-time setup. Finish that and come back.

Step 2: Open chat

Click Chat in the left menu. This is where you’ll talk to any agent — the Admin Agent today, plus any custom agents you or your team set up later.

Step 3: Select the Admin Agent

At the top of the chat window is the agent picker. On a new account, the Admin Agent is selected by default. If you see a different agent, click the picker and choose Admin Agent from the list.

Step 4: Send your first message

Type hello (or anything, really — the Admin Agent is patient) into the box at the bottom and press Enter, or click Send.

Step 5: Watch the reply come in

Within a second or two, the Admin Agent’s reply should start showing up word by word, the way you see AI assistants type in other apps. If yours arrives all at once after a long pause, something’s off with the live connection between your browser and Auxot — see Go deeper → Troubleshooting below.

What’s next

Reference

  • Pages in Auxot: Chat, System Health
  • Keyboard shortcuts: Enter to send, Shift+Enter to start a new line
  • See also: Tutorial 09: Connect a cloud AI model