Approach
From a real case to live operation.
No concept paper that ends up in a drawer. We work on a case from your day-to-day business, at a fixed price and with a defined result, and build from there.
Process
Four steps, each with a clear result.
First call and proposal
30 minutes, without obligation. Within a week a fixed-price proposal, or an honest no if the effort does not pay off.
Architecture and first module
We settle the target picture and deliver one process with real data and a measurable result agreed in advance.
Operation with controls
Tests, logs, alerts and approvals. The software runs, and you can see what it does.
Handover or support
Source code, documentation and training for your team. Ongoing operation for a flat fee if you wish.
Approach
Step by step, each with a result.
Photo: Rubén García on Unsplash
Principles
What you can hold us to.
A person approves
Agents and automations propose, people decide. Nothing goes live without approval.
Source of every figure
Every value can be traced to its source, every run is logged.
No lock-in
What we build for you belongs to you: source code, documentation and access.
Data under control
What goes to an AI model and where it is processed is agreed before the project starts.
Questions
What people often ask.
Do we need AI for this?
Often not. Many processes can be automated with clean code and clear rules, reproducible and cheap to run. We use AI where text has to be understood, for instance when reading contracts or receipts; then we use large language models from established providers, embedded in rules that calculate and check.
What is agent architecture?
The question of how AI agents are wired into your systems: which actions they may take, which data they can reach, where a person approves and how every step is logged. We settle this before an agent goes into production and enforce it technically, for example with a dedicated control server and an allow-list.
What does a project cost?
Within a week of the first call you receive a fixed-price proposal per module with a defined result. For larger system landscapes the Architecture Review comes first, also at a fixed price. Ongoing operation can be covered by a flat fee.
Where is our data stored?
On servers in the EU or in your own infrastructure. Whether and which data goes to an AI model, and in which region it runs, is agreed and documented before the project starts.
Do we own the code?
For work done for you, yes. Source code, documentation and access are handed over. You can continue on your own or with another partner at any time.
What does the EU AI Act mean for us?
Companies using AI must take measures on their staff's AI literacy (Art. 4). Since the Digital Omnibus, Regulation (EU) 2026/1744, the duty is to support the development of that literacy. We document purpose, data flows and human oversight for every automation and train your team to work with it. Legal assessment of individual cases we leave to your trusted advisers.
How are Esox Media and Esox Consult related?
Both are divisions of Esox Consult GmbH in Vienna. Esox Consult organises processes and data for family offices and family holdings; Esox Media designs systems, writes software and automates processes for companies.
When do you advise against automation?
When the process is rare, keeps changing or is already well covered by off-the-shelf software. A custom build then costs more than it saves, and we tell you so before any proposal.
Which process costs you the most time?
Send us two or three sentences about it. In a first call, 30 minutes and without obligation, we work out whether and how it can be solved; a fixed-price proposal follows within a week.