An AI agent for business is a program that carries out several steps of a task on its own, such as reading an inquiry, looking up data and drafting a reply, and asks a person before any important step. I build agents that work with your systems and whose decisions can be checked.

An agent starts by drafting. It may act on its own only where the numbers allow it.

Who it is for

  • Customer support with repeated questions and lookups across systems
  • Document processes where data has to be checked against rules
  • Companies that want AI connected to their CRM, email or internal tools

How it works

  1. We describe the task step by step and decide what the agent may do on its own.
  2. I connect the agent to your systems through APIs or MCP, with only the rights it needs.
  3. I test it on real cases and on attempts to misuse it.
  4. I launch it with approvals and a log of every step.

What you get

Task and limits

What the agent may do on its own, what it only drafts and what it never does.

Tools via API or MCP

The agent can reach only what it needs, with restricted rights.

Shadow mode

The agent runs on real data while a person still does the work, and we compare results.

Abuse testing

Attempts at instructions hidden in documents and e-mails before launch.

Log of every step

What the agent read, what it suggested and who approved it.

In practice

ClaimPilot prepares an insurance claim decision that the claims handler only reviews: in testing it read 99.3 % of critical fields correctly, invented no values, and instructions planted in documents didn't change a single decision.

  • ClaimPilot

    Insurance claims: 99.3 % of critical fields correct.

  • AGL Assistant

    Emails in 6 languages, the team only approves.

What changes in the company

Repetitive tasks get done on their own and people are left with a queue of exceptions instead of a queue of everything. Every decision the agent made can be checked afterwards.

Technology and data security

The agent has only the rights it needs, every step is logged, and amounts, limits and deadlines are calculated by code, not the model.

Price

AI agent prototype €3,600–5,400, 3–6 weeks. Pricing

FAQ

What's the difference between an AI agent and a chatbot?

A chatbot answers questions. An agent also acts: it looks up data in your systems, fills in a form or prepares a document, and waits for approval before any important step.

Can the agent do anything without my consent?

Only what you explicitly allow. I set up sending an email, a payment or a change in a system so that it waits for a person to approve it.

How do you protect the agent from misuse?

I test it against instructions planted in documents and emails. In ClaimPilot, such instructions didn't change a single decision.

How much does an AI project cost?

Private AI over your documents starts from €490, an AI audit costs €540–1,080, automating one process €800–2,400, and production RAG or an AI agent prototype €3,600–5,400. My hourly rate is €35–45 and the first 30-minute call is free.

Related