Nicolas Fonteyne
Founder
Quantitative analysis, financial markets, software and entrepreneurship.
LinkedInWhy I built Orion
My career has run through quantitative analysis, financial markets, software and entrepreneurship. In all of those worlds I found the same problem: the information that decides the day exists, but it's scattered across conversations, spreadsheets and inboxes, and pulling it together costs someone hours.
Orion started as a tool for answering questions about a company's data. Talking to small business owners, it became clear that the most important data never even reaches the spreadsheet: it's in the customer's voice note, in the supplier's email and in the photo of the delivery receipt.
That's why Orion became a framework. Instead of asking the team to switch tools, we build each company an agent that works with its people, in the channels they already use, and keeps a record of how the operation really runs.
For the first time, technology will bend to the company, and not the other way around.
This is my long-term view of how AI changes software and the companies that depend on it. Orion is the way to start building it now.
Before · Off-the-shelf software
10 disconnected tools
Buy and adapt
The company adapted to the software.
Software was slow and expensive to build, so software companies made generic tools for everyone. Each company had to bend its own way of working around them.
Now · The AI era
One suite for each company
Custom-built
Now the software adapts to the company.
- Bank
- Payments
The company's agent infrastructure
- Agent loop
- Business rules and data
- Private cloud
- Event base
- Accounting
- Carriers
Before, only large companies had something like this. Now small and mid-sized ones can too, with everything connected to the services around them.
Most companies don't have a technical team, or don't know what AI can already do today. We design and implement it for them.
So why not a ready-made product that everyone downloads, installs and starts using?
Because that would be going back to the old model. A product that's the same for everyone is, once again, a tool built for the average company, and whoever uses it ends up adapting to it. But every operation has its own orders, rules, sources of information and channels.
Starting from scratch with every customer doesn't make sense either. That's why Orion is a framework: a common foundation, with everything every agent needs, on which we design and implement each company's agent.
What belongs to each company
- The agent's name
- The business rules
- The knowledge sources
- The tools
- The team's channels
What Orion brings to every agent
- Listening to voice notes and reading photos and PDFs
- Using the company's tools
- An event base for every step
- A person's approval on what matters
- A private cloud for the data
The agent belongs to the company and carries its name. Orion sits underneath, and because each new agent starts from what already exists, it's ready faster and costs less than the last one.
- 01
For decades, the company adapted to the software.
Building software was hard, slow and expensive. That's why software companies emerged, with generic tools like CRM and ERP, made to serve everyone. SaaS won the information age because it scales: you write the code once and sell it to everyone.
The price was customization. Whoever buys a fixed solution adjusts their own way of working to it, and ends up with ten tools from ten vendors that were never designed together.
- 02
AI has collapsed the cost of building custom software.
A software suite built for a single company has always been possible, but only the giants could afford it. With AI, the cost of developing custom solutions has plummeted.
I believe this changes what companies expect from software: fewer and fewer loose systems, more and more a suite of their own, dedicated to the way they actually operate.
- 03
Every company will want to own the software that governs it.
The software that decides how orders, purchases and payments happen is the center of the operation. As custom software becomes viable, companies will buy fewer generic off-the-shelf solutions and will want a system designed for them, one that follows their rules.
Banks, payment providers, accounting and carriers remain third parties. The difference is that they plug into a suite that belongs to the company and follows its rules.
- 04
Small and mid-sized companies are the ones who need help the most.
They are most of the economy, and their information lives scattered: customers write by email, the team talks on WhatsApp or iMessage, and suppliers join the same exchange. Most don't have a technical team to put together a suite of their own, and many people don't even know what AI already allows today.
The cost of building has fallen, but the ability to take advantage of that hasn't spread yet. It's one of the reasons the gains from AI don't show up in GDP yet, and it's in that gap that Orion works.
- 43.5%
- of U.S. GDP comes from small businesses.
- 1 in 5
- American companies already use AI.
- < 20%
- among those with up to 4 employees, versus 37% among those with 250 or more.
Census BTOS, May 2026 · SBA Office of Advocacy, 2026
- 05
Before the agent comes the foundation.
For AI to work in a company, it needs a single, reliable source of what's happening: which orders are open, from which customers, which were delivered and which came back. As long as that answer is scattered across inboxes and conversations, no agent can act. The barrier isn't the agent, it's everything underneath it.
With the foundation in place, the agent follows the channels the company already uses and comes to know everything that happens in it, without anyone having to feed it. The person remains the one who decides.
Every company with its own software suite, custom-built and built with help. That's what Orion exists for.
We design and implement custom AI agents so that small and mid-sized companies can capture the real value of AI, something that until now only large ones could have.
Tell us how your team works.
We design your company's agent, with the name, rules and tools of your operation.
What to tell us
A few points that help us understand your company better. You don't need to cover them all.
- What the company does and how many people work in it
- Which areas exist: sales, purchasing, logistics, finance…
- Where information travels today: WhatsApp, email, spreadsheets, ERP, voice notes
- Where the team loses the most time or where things get lost
- Repetitive work that could be automated
- What you, as a manager, would like to follow without having to ask