What is an AI agent? Key concepts (model, tone, tools, budget)

Understand in minutes what an AI agent is in Omnifox and the four key concepts you'll configure: model, tone, tools and budget.

Jun 26, 2026

Watch this step in the video · The AI Agents screen

What is an AI agent

An AI agent in Omnifox is an automated assistant that talks to your contacts over chat and calls. It answers questions, qualifies leads, books appointments and runs actions (assign, close, tag) without a human typing every reply. You set it up once and it works 24/7 within the rules you give it.

Before you build or edit one, it helps to understand the four concepts you'll touch the most.

Model

The model is the AI engine that generates the replies. Omnifox manages it for you: you don't pick a provider or pay a separate API. Every reply consumes AI tokens from your plan's monthly allowance (for example, Lite includes 30,000 tokens, about 600 replies); once it runs out, the AI keeps going on your Omnifox balance (top-up wallet). More conversations and longer answers use more tokens, which is why you'll later find length controls.

Tone

The tone is the personality the agent replies with: friendly, formal, sales-driven, and so on. You shape it in two places:

  • The Instructions tab, where you write the prompt and describe how it should speak.
  • Behavior → Reply style, where you choose between Fixed tone (always keeps the prompt's style) and Adaptive tone (mirrors the customer: formal if they're formal, emojis if they use emojis).

The same tab has Temperature, from Deterministic (same answer every time, great for support) to Creative.

Tools

Tools (or actions) are what the agent can do, not just say. You enable them in the Custom actions tab. There are two kinds:

  • Native: assign the conversation to an agent or team, close it, add tags, transfer to a human, trigger a workflow, book an appointment, and more.
  • Custom HTTP actions (from the Impulsa plan): calls to your own systems (look up an order, open a ticket) through a URL.

Only the tools you turn on are available to the agent.

Budget

AI spend is controlled by your plan's monthly token allowance and your Omnifox balance, under Settings → Billing. Under Behavior you'll still see a Daily token budget field, but it no longer stops the agent: it's disabled across the platform.

Where to find it in Omnifox

  1. Go to Settings → AI Agents.
  2. Create one with Create AI agent or open an existing one.
  3. Walk through the tabs: Identity, Instructions, Custom actions, Knowledge, Voice, Behavior and Sandbox.
  4. Test your changes in Sandbox before publishing.

Verify

If the agent replies with the expected tone and runs the actions you enabled in Sandbox, your foundation is ready.

If something feels off

  • It won't use an action: make sure it's enabled in Custom actions and described in the prompt.
  • Replies are odd or too long: lower the Temperature and tighten the token limit under Behavior.
  • It went silent: check your AI allowance and Omnifox balance under Settings → Billing, and the agent's active hours.

Tips

  • Start with a fixed tone and low temperature; add creativity only if you need it.
  • Enable only the tools the agent will actually use: fewer options means sharper decisions.

Related

Was this helpful?

Related articles

Message us on WhatsApp