"AI" itself is the easy part: a subscription, a credit card, five minutes. The hard part is the rest — which happens to be your week: evenings copying orders from the portal into the management software, the quote that takes half an hour of copy-and-paste, the info@ inbox where requests get lost. The right question isn't "which AI should I buy" but "which hours of my week go on work a machine would do just the same". That's the starting point: your processes, not the catalogue of whoever sells magic.
The service covers the whole chain: we map the repetitive processes, automate the flows with proportionate tools — n8n, Make, scripts, the APIs of the software you already run —, make documents fill themselves in from data you've already typed once, and add a tailored AI assistant only where it genuinely earns its place, integrated with your existing management software. A delivered automation means something precise: a flow that runs, with written documentation of what it does, where it runs and how to switch it off — the red button is part of the handover — plus a human check on the results, agreed together rather than left to chance.
We start small, with numbers on the table: half a day mapping your real processes — charged by the hour like any other single job, quote by email first —, then a fixed-price quote for the flow with the best ratio of hours saved to cost, and only later, if it makes sense, a contract to maintain the flows. Two honest warnings come first: not everything should be automated — a broken process, automated, is still broken, just faster — and we are not a software house: the management software stays yours; we make it work harder.
The quote comes from the mapping, not from a price list. The method has three steps. First: half a day on your real processes — who does what, in which programs, how many times a week — closed, like every job of ours, with a written report: the processes ranked by payback, the hours they eat, where it pays to start. Second: we automate the flow with the best ratio of hours saved to cost. Third: we measure for a few weeks, and only if the numbers hold do we widen to the next process. No six-month grand project: one flow at a time. It's the method we use on ourselves — the lab itself runs on automations built in-house: quotes, reminders, filing — and it's why we never propose anything we wouldn't run.
The right tool is the proportionate one: subscribing to ten platforms is not a strategy. Before adding software we look at what's already there: often your management software plus one well-built flow connecting it to the rest is enough. Where an engine is needed we use working tools — n8n, Make, a script, the APIs of the programs you already pay for, the doors they use to talk to each other — chosen on cost, confidentiality and ease of maintenance, not on fashion. This is where the robot-work disappears: orders that travel from the portal into the management software on their own, reminders that go out on the right day, and automatic documents — quotes, delivery notes, covering letters — that fill themselves in from data typed once, instead of being rebuilt from copy-and-paste every time.
An AI assistant is useful when it knows your business and knows its place. No magic, says the least suspect source: the EU regulation on artificial intelligence, in its own recitals, writes that "A key characteristic of AI systems is their capability to infer" — a capability that "refers to the process of obtaining the outputs, such as predictions, content, recommendations, or decisions". Inference: the system deduces, it doesn't know. Pointed at your own material — price lists, datasheets, procedures — an assistant built on your documents does genuinely useful work: drafts replies for the info@ inbox, finds the right datasheet in seconds, summarises the paper trail of a job. And a house rule: an assistant that talks to the public introduces itself for what it is — a machine, with a person behind it. Be wary of anyone selling AI as magic without ever having seen your processes.
Generative AI gets things wrong: that's why human control over the results is a line in the project, not a footnote. In every flow we decide together, in writing, what runs on its own and what waits for your click: an internal reminder can go out by itself; a quote, a price, a promise to a customer passes a human eye first. And the oldest rule of the trade still holds: a broken process, automated, is a broken process running faster — mapping also exists to tell you what to fix first, and what not to automate at all. At handover every automation comes documented: what it does, where it runs, how to switch it off — the red button is part of the delivery, not a favour. And the accounts and keys behind the flows belong to your business: written down and handed over, never kept as leverage.
The management software you already run is the centre of the project, not the obstacle. Integration starts there: with documented APIs the flows talk to it directly; without them, it's exports and imports — and if the vendor keeps the doors shut, we tell you at the quote. On data the rule is written first: which information leaves towards external services is agreed in black and white, and wherever possible the flows run in-house — the GDPR, in its recitals, wants personal data "adequate, relevant and limited to what is necessary for the purposes for which they are processed", and a well-built automation moves less data around, not more. Two final truths: a flow is as alive as the IT underneath it — server, network, updates: the trade of IT support — and the data it produces is worth as much as the copy protecting it: the backup and continuity chapter.
| Processes on the table | how many there are, how often they run and how many hands they touch: mapping ranks them by payback |
|---|---|
| Systems to connect | management software, portals, spreadsheets, mailboxes: every extra system is a bridge to build and test |
| Open or closed doors | documented APIs make a flow straightforward; without them it's exports and imports, and the hours change |
| Where the flows run | on a machine on your premises or in the cloud: costs, data confidentiality and upkeep all change |
| How much human control | what runs on its own and what waits for your click: every approval step is designed, and costed |
| Keeping flows alive | portals and programs change without warning: you choose whether to watch them yourself or under a contract |
What genuinely moves the quote for this service: how many workstations and people, what is already in place, what must stay up. Agreed together before we start — and the quote, numbers included, arrives by email: one-off jobs by the hour, ongoing support priced to fit.
ElectrONIK Lab is an independent workshop: it is not an authorised service centre and is not affiliated with the manufacturers. The trademarks named here belong to their respective owners and are used only to identify the devices we work on. Your statutory consumer rights towards whoever carries out the repair are unaffected.
A free consultation, a clear quote up front and a written report after every job.