A catalogue instead of scattered lists
Every entry with purpose, trigger, boundary, owner, subject expert, connected systems and task areas. Grouped by department or by the system it uses — and a search that covers all of it at once.
Agentario is the inventory for a company's AI agents and automations: who runs them, what they do, how far along they really are — from the first idea to established operation.
You host it on your own server — your entries never leave your company. No price list, no sales pitch.
In most companies, nobody knows what is already running.
Every department tries something out. It gets recorded in chats, notes and people's heads — never in one place.
“It's already running” means something different in every department: a prototype on a laptop in one, live operation with real customers in the next.
Who is responsible for an agent usually becomes clear on the day it breaks.
Every entry with purpose, trigger, boundary, owner, subject expert, connected systems and task areas. Grouped by department or by the system it uses — and a search that covers all of it at once.
Five stages from idea to established operation, each with a definition and a test question. Planned and productive sit side by side in one list — filtered rather than separated, so nobody keeps two versions of the truth.
“Runs reliably, is used routinely, has an active owner and demonstrable value.” — the definition lives in the app, not in someone's head.
Three roles: anyone may read, users maintain, admins decide. Whoever wants to take something into production files a request — and which fields have to be filled in first is up to you. Every decision lands in the log.
An agent that uses others forms a chain. Agentario derives the levels from that by itself, shows them as a list or as a map with a focus mode — and warns you before a chain gets so deep that nobody can follow it.
Metrics, maturity distribution, coverage per department, activity over 30, 90 or 360 days — as a finished management report built from blocks you pick. It runs on your own AI key; without a key the feature simply stays off.
Two views from a sample installation with 42 entries across eleven departments.
Dashboard. How much is running, how far along it is, which department is still missing.
Hierarchy. Which agent uses which — across department boundaries too.
No tiers, no enterprise edition. This is the whole app.
Agentario runs on your own server. You get the application itself, not access to mine.
A couple of sentences about what you have in mind and how you work today. That is all it takes.
As a repository or as a ZIP archive, whichever suits you. With setup instructions.
On your server, with your data. I never see your entries — and we need no contract about it.
There is a self-serving reason: I want to see how other companies handle AI agents — what questions they ask, where it gets stuck, what they are missing. For that I need people who actually use it.
If you run Agentario and tell me now and then what works and what does not, that is worth more to me than an invoice. I reply personally, with no sales team in between.
Write me a few sentences about what you have in mind and how you work today — that's all it takes.
Write to mehello@borishaggenmueller.de — I use your details only to reply to you.