We keep your agents working, and we answer for it when they don't
AI projects rarely fail on launch day. They fail in month four. The vendor changes the model version, output quality slips and nobody notices, and the person who built it has moved on. Your team quietly goes back to doing the work by hand, while you keep paying for both.
We run your agents in production under a service level agreement. We're contractually responsible for them working, not just for delivering them.
Trusted since 2004 by the public sector, energy and heavy industry: buyers who audit their suppliers before they sign. See the case studies.
What you get
It keeps working
- Model versions pinned, upgraded only after testing on your reference cases
- Output quality monitored, not just server uptime
- Alerts when the escalation rate jumps, a sign your data has changed
- Running costs optimised at our expense
- A fallback if a model vendor goes down
Someone answers for it
- A named contact who knows your setup
- Response times by severity, written into the contract
- A monthly report your board can read
- A full decision log, kept and searchable
- A quarterly compliance pack, ready for an auditor
What the SLA commits us to
Availability. Incident response times by severity. A minimum output quality, below which we investigate at our cost. How often we report. When your compliance evidence arrives.
All of it is measurable. If we miss a target, it shows up in your own monthly report. You don't have to catch us out.
Your people stay in charge
Each agent has a confidence threshold. Above it, the agent acts. Below it, a named person on your side gets the case with the reasoning attached. We set the threshold during the pilot and tighten it as history builds.
Your staff work only on the exceptions, where their judgement is worth paying for. It's also the human oversight the EU AI Act expects.
What's yours, and what stays ours
The agent framework, the prompts and the reusable parts behind them are our product. They're why we can do this faster and cheaper than an in-house build.
Always yours:
- Your data. Exportable in a standard format at any time. We don't resell it or train shared models on it.
- Your audit trail. Every decision log, kept and exportable. It's your evidence under the AI Act and NIS2, so we never hold it back.
- Your system documentation. What the agent does, what data it touches and how oversight works. Everything your AI register needs.
- Your business rules, written down properly. Usually for the first time.
Source code escrow is available on enterprise contracts. Code, prompts and configuration sit with an independent third party. They're released to you automatically if we become insolvent, seriously breach the contract or stop the service. NIS2 asks you to assess supplier dependency, and escrow is a clean answer to that.
What it costs
A fixed monthly fee. The same amount every month, whatever the agent processes.
We don't bill per token, per document or per model call. Model usage is in the fee and it's our problem to optimise. A busy month doesn't change your invoice, so you can put the number in the budget in January and forget it.
We set the fee after the pilot, once we know what the process involves. Contracts usually start at 12 months.
One agent first, not five
We won't take on five processes at once, even if you ask. One agent, one process, running next to the manual work until the numbers hold. Then we expand. Tell us on a call what you'd want running, and we'll tell you what it takes to keep it running.
Common questions
Why outsource operations rather than just the build?
Because the hard part starts after launch: models change, vendors deprecate versions, data drifts, and documentation obligations are continuous. Few companies have someone doing this daily.
What is in the SLA?
Availability, incident response times, output quality thresholds, monthly reporting, and a quarterly compliance evidence pack you can put in front of an auditor.
What if we want to stop or change supplier?
Your data, your audit trail and your system documentation are yours and can be exported at any time. Without them you could not prove compliance, so we never make them conditional. The agent framework and prompt engineering remain our product. For enterprise contracts we offer source code escrow: code and configuration sit with an independent third party and are released to you automatically if we become insolvent, materially breach the contract, or discontinue the service.
How is this billed? Do we pay per token or per document?
No. A fixed monthly fee, the same every month regardless of volume processed. Model consumption is included and is our problem, not yours. You can put the number in the budget at the start of the year and stop thinking about it.
Find out in 30 minutes if it's worth doing.
Tell us about one process that eats your team's week. We'll tell you on the call whether an agent can take it over, roughly what it costs and what it saves. If it's not a fit, we'll say so.