The Embedded Week: how Arclight builds custom AI

The Embedded Week is how Arclight builds custom AI. We sit with your team for a week, watch how work really gets done, map where the time goes, then build a solution made for your company, not a generic product. After that we train your team to run it.

Arclight Team
7 min read

What is the Embedded Week?

Most AI projects start with a tool and then go looking for a problem. The Embedded Week works the other way round. We sit with your team for a week and watch how the work really gets done: the screens people open, the hand-offs between them, the copy-pasting from one system to another, the waiting for a reply or an approval. Nobody can describe all of that from memory, which is why we come and look.

From what we see, we map where the time goes and which tasks are worth changing first. Then we build a custom AI solution for your company specifically, not a generic product with your logo on it. It works with your tools, your documents and your rules. Finally, we train your team to run it, so the people who use it understand it and can ask for changes with confidence. Each step feeds the next, and nothing is built before the map exists.

We work this way because the value of AI sits in the details of how a particular company operates. A solution that fits those details gets used. One designed for an average company and then adjusted usually does not.

What happens in the week, day by day?

The week is simple on paper. On day one we meet the team and agree which parts of the business we will look at. On days two and three we sit beside people as they work and ask why at each step. For examples of what we tend to find, see our guides to AI automations by department and, to see how the same thinking applies inside one specialised sector, our guide to AI for insurance brokers. On day four we sort the findings by value and effort. On day five we present the map and a build plan to the people who will use the solution.

After the week, the build begins. We use Arclight's services only where the map shows they pay off, and nothing is added for its own sake. The usual pieces are automation, AI apps, AI agents and training, used together or alone. Each one starts from the business problem, because a technical choice also commits a company to a budget, a team and a way of working.

Throughout, your team stays in charge. The people who do the work check what we find, so the map reflects reality and not our assumptions.

What do you get from the week?

A method only counts if you can check it. So here is what you take away, and you can read more on the company page, which applies to every project.

First, a map of the work: where time goes, which hand-offs break, which tasks repeat and what each one costs your team in effort. Second, a build plan that ranks what to automate first, with the tools, data and approvals involved. Third, once you agree the plan, a custom solution built for your company, followed by training so your team can run it and change it. Nothing in the map is a template, because nothing about your company is.

We also give you an opinion. If an existing tool is enough for part of the problem, we will say so, and the plan will use it. Custom work is kept for the needs that justify it, with human review wherever a mistake would be costly. Every solution comes with training, so your people understand what they are using, know when to trust it and know when to check it.

What do we look for while we watch?

We look for skilled people spending time on work that needs no skill. Tools that do not talk to each other, so someone retypes the data. Copy-pasting between documents, emails and spreadsheets. Waiting for an approval, a reply, a file or the one colleague who knows the answer. Each of these is a candidate for AI, and each is invisible in a meeting room, which is why we watch the work instead of asking people to describe it.

We also note what should stay human: judgement calls, sensitive conversations and anything where a mistake would be costly. The aim is to remove the drudgery, not the craft.

Who for?

The Embedded Week suits companies whose work spans many tools and hand-offs. Our pages for permanent staff and Maison Kameur each show how it applies. If you want AI built around how your team really works, book a meeting.

The Embedded Week: your questions

Answers to the questions people ask us most about the week.

How long is the Embedded Week?

A week. We are with your team for that time, watching, asking and mapping. It ends with a map of how the work really happens and a build plan. The custom build and the training follow from that plan.

Who needs to be involved in the week?

The people who do the work, plus someone with authority to approve changes. We need their time to watch and ask questions, not long meetings or documents prepared in advance.

Is the result a standard product?

No. Each solution is built for your company, around your tools, data and rules. We do not resell a generic product with your logo on it.

What do we have at the end of it?

A map of the work, a build plan, then a custom solution and training so your team can run it.

How do we book an Embedded Week?

Book a meeting using the link on this page. We will talk through your team and how work is organised, then plan the week.

Tell us what you want to automate.