Operation & Ongoing Care
An agent that has been set up once does not stay useful by itself. Interfaces change, models get retired, price lists and workflows go stale, and without ongoing AI agent care these small drifts add up to real errors. Operation & Ongoing Care is the monthly module built for exactly that: the ongoing maintenance of your AI agents, monitored, documented, and backed by one named contact instead of a ticket system. That way you do not become the fire brigade the moment something breaks, and the first project turns into a managed AI service that grows with you month by month.
Who this is for
- You have one or more agents in production and do not want to find out one has stopped working only when a customer complains about a wrong answer.
- Your team has neither the time nor the expertise to keep models, prompts and connections maintained long term, that was already tight during setup.
- You want to move a step further every month rather than stalling after the first project and leaving the agent exactly as it was at launch.
- You want to be able to show your management or team internally what the agents deliver, with numbers rather than a good feeling.
- The colleague who originally helped set up the agent has left the company, and nobody else understands the prompts, connections or knowledge base behind it.
- You have introduced a price change, a new location or a new product in the last few months, and the agent is still giving the old answer.
What the module concretely delivers
Monitoring in the background
We keep a continuous eye on your agents and catch failed runs, timeouts and changed interfaces before your team or your customers notice them. If a run fails repeatedly, we step in rather than waiting for someone to report it.
Monthly report
A short, readable overview of what the agents took over that month, how often they handed back to a human, and where things got stuck. That gives you a basis to show the value internally instead of just claiming it.
Company knowledge stays current
New prices, changed templates, different workflows: we update the knowledge base your agents draw on. Outdated company knowledge is the most common reason for wrong answers, and that is exactly what ongoing maintenance prevents.
One expansion step per month
Every month adds one improvement, whether an additional workflow, a new connection, or an extension of an existing agent. We agree the priority together, so the expansion happens where it delivers the most.
One named contact
You reach the same person rather than a ticket system or a rotating support team, with a committed response time within one working day when something breaks. That person knows your agent, your workflows and your company knowledge, so you never have to re-explain the context.
Model changes without disruption
When a provider retires a model, or a new version performs clearly better, we test the switch and roll it out in a controlled way instead of your agent suddenly behaving differently overnight. You hear what is changing beforehand, not only once it becomes noticeable.
How the ongoing care works
Handover
We take over the existing agents, review the prompts, connections and knowledge base, and set up monitoring. Together we agree what we may change on our own and what needs your approval.
Monthly rhythm
Secure operation, send the report, agree the next expansion step: usually one short meeting a month is enough. Urgent fixes do not wait for that meeting, they are handled immediately.
Escalation on incidents
If an agent fails or repeatedly gives wrong answers, we get in touch within one working day and fix the root cause, not just the individual case. For critical workflows we agree in advance who gets informed if something serious happens.
Cancellable at any time
Cancellable at the end of any month without a reason; the agents, the company knowledge and the access stay with you, handed over documented. There is no licence that expires with the cancellation, and no data export you have to fight for.
Pricing
Operation & Ongoing Care is billed as a monthly flat fee. The amount depends on how many agents are under care and how much expansion is planned per month. Hosting and model costs are itemized separately so it stays visible what is care and what fluctuates with usage. We set the concrete figure in the free initial consultation.
Cancellable monthly, no minimum term. All amounts excl. VAT.
See all servicesFrequently asked questions about Operation & Ongoing Care
Yes, a handed-over agent works initially without ongoing care as well. The module catches what happens afterwards: changed interfaces, new model versions and outdated company knowledge, which lead to errors over time.
The agents, your company knowledge and all access stay with you. We document the current state and hand it over to your team or to another provider. There is no licence that expires with the cancellation.
That depends on the agreed scope: a smaller extension at the lower end, continuous expansion across several workflows at the upper end. What is realistic gets clarified at handover.
No, they are itemized separately so it stays visible what is fixed and what fluctuates with usage. On request we handle the billing, or the costs run directly through your own accounts.
Yes, this module only makes sense once at least one agent is running in production. Without existing operation there is nothing to monitor or expand. If you do not have one yet, we say so openly in the initial consultation.
It depends on how many agents are under care and how substantial the monthly expansion step is: one agent with small monthly tweaks costs noticeably less than several agents with continuous expansion. Hosting and model costs come on top separately and fluctuate with usage. We set the concrete figure in the free initial consultation, excl. VAT. There is no minimum term, so you only pay for as long as the module is worth it to you.
For incidents we get in touch within one working day, in practice often much faster because monitoring frequently catches the issue before you notice it. For critical workflows we agree in the initial consultation whether a shorter response time makes sense.
Yes, that is possible. We first review the existing setup to understand how it works before we put monitoring in place. For very unclear or undocumented systems, that can require a short assessment upfront.
