Struan builds and runs AI employees as a managed service, one scoped role at a time. Each role draws on skills for finance, admin, operations, writing, data and customer support. A UK business starts by naming the recurring work, the required output and the person who approves it. The job description then records the sources and tools the role may open.
Find the work that comes round every week
Look at where your team spends the most time on repetitive work with clear inputs and outputs: routine customer enquiries, invoice processing, scheduling and calendar management, content drafting, or data entry. The job that consumes the most repeated time is usually the strongest candidate for a first role, whether that sits in finance, admin, writing, data or support.
Name only the tools the role needs to open
List the tools the role would need to touch, such as your CRM, accounting software, email or calendar. Struan does not assume a connection to any named system in advance. The connection is scoped in the job description and proved on your own data before the role runs, so your team is not relying on a list of supported software published ahead of time.
Choose one role before you choose five
Start with a single role rather than several at once. A wide first deployment makes it harder to judge what changed and why. The job description sets out exactly what the role covers and what it does not. Adding a second role, or extending the first, happens only when you and Struan agree a change in writing, not through the role expanding on its own.
What procurement should expect: cost and control
The cost of building the role sits apart from the ongoing managed-service fee. Both are agreed in writing before anything is built, and the fee covers running the role, tuning it and the weekly sign-off. One instruction from your team stops the role, and a written stop procedure applies whenever that happens.
How the first role actually gets built
The process runs in stages. A free scope call covers the jobs your team repeats, the tools you run and who signs off what, and if a role does not suit the work, you are told on the call. You then agree a written job description naming the jobs, the sources it may open, what comes back and who approves it. Struan proves the role on your own live work before you agree to run it, so you see real output before committing. Once live, the role works inside the chat tool your team already uses, such as Microsoft Teams or Google Workspace, and a named person at Struan signs a report your team reads every week. The full process is set out on the managed-service process page.
Questions worth asking before you commit
Do we need technical skill in-house?
No named technical hire is required. Struan builds and connects the role during setup, and your team's job is naming the work and approving what comes back.
What happens if the AI employee gets something wrong?
Struan corrects the mistake and tunes the role. A named person at Struan signs the weekly report your team reads, so there is someone accountable for the run.
How quickly will we see a working role?
There is no fixed timeline. The pace depends on how clear the job description is and how quickly you approve the proof stage. Nothing goes live until you have agreed to it in writing.
If a role looks like a fit, book a scope call to name the jobs worth moving. If you want to see real work first, existing deployments are written up on the proof page. Two further explainers cover what AI as a hire means and how managed AI employees are run day to day. You can also get in touch directly if you would rather talk it through first.