Managed AI operations
We keep
live AI healthy
Managed AI operations is for a software company with AI already in production, whether we built it or we inherited it. The monthly agreement keeps it healthy: evaluations, regression, drift, latency and cost, incident response, and a monthly report. New features are fixed-price change orders.
The facts
At a glance.
- Who it is for
- A software company with AI already in production, built here or inherited.
- Length
- A monthly agreement.
- What you get
- Evaluations, regression, drift, latency and cost, incident response, and a monthly report.
- What happens after
- New features are fixed-price change orders.
- Pricing
- Contact us for pricing.
Something already live
Tell us which AI nobody owns
A system we built, or one you inherited. We say what the monthly care includes, and what the first fixed-price order would be if the evaluation set is missing.
In the agreement
What the retainer covers.
Care of systems that are already live, whether we built them or we inherited them.
Evaluations on a cadence
The evaluation set runs on a schedule, not only when someone remembers to check.
Regression on model and prompt changes
A change that fails the set does not go out. The gate sits in continuous integration.
Drift, latency and cost
Alerts when answers move, when it gets slow, or when the bill moves.
Incident response
A person from us when the system is wrong in production. We write down what happened and what we changed.
A monthly report
What ran, what failed the evaluation set, what it cost, and what we recommend next. The next feature is not started from the report until you approve a price.
How it works
How the agreement works
Same commercial shape as Platform Partner. The retainer is care. Features are separate.
Start from the system you have
Aidoni-built or inherited. We learn the prompts, the data boundary and the evaluation set. If there is no evaluation set, writing one is the first fixed-price order.
Monthly care
Evaluations, regression, monitoring, incidents and the report. Small fixes that keep it healthy sit inside the agreement. We do not start new features from the retainer.
New features are change orders
A new tool, a new source or a new workflow gets a fixed price before we start. Nothing outside the agreement until you have approved it.
The retainer
is care.
Features are separate.
Care, with a clear edge
Care, with a clear edge.
You should know what the monthly agreement includes, and what it does not.
The split is explicit
The retainer keeps it healthy. New features are fixed-price change orders. Same language as Platform Partner.
Evaluations from the start
If the system has no evaluation set, that set is the first thing we put in place. Later changes have to pass it.
Your environment
We operate inside the boundary you already have, on Amazon Web Services, Microsoft Azure, Google Cloud, GleSYS, or on your own servers. Data stays there. Audit logs stay on. We do not train on your data unless you ask us to, in writing.
The platform, the sprint, the team
Where this sits with the rest.
Managed AI operations is the care package. The others are how a system gets built, or how a person sits in your team.
Platform Partner
Takeover of an app, portal or connected product: platform check, monthly care, then fixed-price AI builds.
Read more →Fixed projectAI production sprint
When the AI is not live yet. One system in four to six weeks, then this agreement if you want us to keep it.
Read more →Engineering teamsAI coding rollout
When the live system is how your engineers use coding agents, and the gap is the team setup.
Read more →ConsultingForward-deployed engineers
A senior in your team when care is not enough and you want someone building with you.
Read more →Private modelPrivate model / fine-tuning
An open model tuned on your own data, run where your data must stay, and only if an evaluation says the hosted model falls short.
Read more →AIAll AI packages
How the sprint, this retainer, AI coding rollout and Private model / fine-tuning fit together.
Read more →Related insights
Keeping AI in production.
What has to be true before a system should be scaled, and how to watch it.
Agent readiness before you scale AI
Agentic workflows fail fast when data access, ownership, and logging are unclear. A short readiness sprint surfaces the gaps before production.
Read insight →AI Product Development · 8 min readAI pilot to production: a practical playbook for Nordic product teams
Most AI pilots never become product value. Here is a field-tested playbook for taking an AI initiative from demo to stable production impact.
Read insight →Questions
Common questions.
The same questions we get about Platform Partner, scoped to AI.
What is in the retainer, and what is a change order?
The retainer keeps the system healthy: evaluations, regression on model and prompt changes, drift, latency and cost, incident response and a monthly report. A new feature, a new data source or a new workflow is a change order at a fixed price, approved before we start.
What does it cost?
A monthly agreement, priced before we start. We do not publish the amount. It follows how many systems are in care and how they are hosted. Change orders are priced one by one. Contact us for pricing.
Does Aidoni have to have built it?
No. Inherited systems are a normal start. If what you have is an app, a portal or a connected product with no owner, start with Platform Partner. This page is the AI on top of a system.
What if there is no evaluation set?
Then the first order is to write the evaluation set and put the continuous integration gate in. We do not pretend to operate a system we cannot tell has got worse.
Where does the data stay?
In your environment, on Amazon Web Services, Microsoft Azure, Google Cloud, GleSYS, or on your own servers, with audit logs. We do not train on your data unless you ask us to, in writing.
Ready to talk?
Bring the system, not a maturity model
We will tell you straight if it needs care, a sprint, or a platform takeover first.
Ready to talk?
Let's talk about
what you're building
Or reach us directly at hello@aidoni.se
