// AI agents & applications
An agent is a role. An application is where they work.
A job to do, access to the systems, and responsibility for the outcome. Several working together cover a whole process — and the application is where they and your people work side by side.
Everyone has a slide about AI. Nobody has a map.
Most companies are on the first two and want to be further; the gap is knowing what is realistic. Two things change as you go up — agents get more autonomous, systems get broader. And each band needs more underneath it, which is what actually decides the cost.
Level
What it does for you
Productivity
Who builds
01
AI chat / Copilot
Answers questions. You still do all the work.Most companies are here
Personal
You — we teach you how
02
Light agent
Performs one task, when asked.Most companies are here
03
Knowledge agent
Knows your data. Answers from your own numbers and definitions.
Team / process
Either — we build or teach
04
Action agent
Does the work, in your systems. You decide how far it may go.
05
Multi-agent system
Runs a whole process end to end, the way a team would.
Business
Us — you keep building on it
06
Agentic application
An application your people work in every day — not a tool they visit.
07
Agentic system
Several applications across the operation, largely running themselves. You set direction.
Levels 01–02 need nothing underneath. Levels 03–04 need part of your data connected and defined; levels 05–07 need all of it — and once that is there, your own people keep building on top.
Knowing your data makes an agent useful. Acting in your systems is what makes it pay.
What a team looks like when part of it isn't human
One team: an architect designing what sits underneath — the data, the definitions and the limits — a manager routing the work, and specialists that each take the repeating part of one job. The decisions stay with your people. These are examples: we build the roles your operation needs.
// 06–07 · Across all of it — agentic applications & systems
Aineo Agency
Architect
Someone decides what these do, what they may touch, and what they must never do on their own — and keeps deciding as the operation changes. That part does not get delegated.
Yours, not ours: it runs in your environment, on your identity and your storage.
// 05 · Domain manager — multi-agent system
Neo
Manager
Routes the work between the others and keeps them consistent — the part that stops ten specialists becoming ten more silos.
A full application usually runs several of these — one per process or function, each with its own manager and its own specialists.
// 03–04 · Domain specialists — AI agents
Jesper
Operative purchasing
Sofia
Supplier performance
Julia
Quality & deviations
Anna
Forecast & S&OP
Emma
Inventory & replenishment
Daniel
Master data
Linda
KPI & reporting
Lisa
Documents & contracts
Maria
Sales & margin
Oliver
Technical support
This team is drawn from manufacturing and supply chain. Project businesses — construction, shipbuilding, engineering — get a different set: schedule, cost to complete, change and risk. These are examples, not an inventory.
// 01–02 · Where most companies are — AI chat, Copilot, and light agents
Copilot
What you already have
It reads your email, your SharePoint files and your Teams conversations, and it answers well from them. Ask what your on-time delivery was last month and it has nothing to work from: it cannot see the ERP, the quality system or the project data. That is usually why the licences are not used to the full.
Everything above level 02 needs the same thing underneath, and that is the part we build: your systems connected, and every number defined once. Ask again and the answer is there — the same one the rest of the company is working from. The licence you already pay for becomes worth having, and Copilot works alongside the agents we build.
Which role would you hire first?
Tell us the process that costs you the most time. The next day you see an agent doing part of it, built on your business.