Skip to content

The stack

What the work is built on.

Eleven pieces, and what each one is actually for. None of them is a wall you would have to climb to leave, and that is a design decision rather than a happy accident.

What we build with

Showing 11 of 11 tools.

  • Engineering

    Custom build

    For the jobs where nothing off the shelf fits the shape of the work.

    Most automations are assembled from the tools below. Some are not, because the process is genuinely unusual. Those get written, and they are yours: the repository can be transferred on the day you ask for it.

  • Intelligence

    Frontier models

    The part that reads, decides and writes rather than following a rule.

    Used where the work needs a reading rather than a lookup: classifying an enquiry, pulling values off a photographed sheet, drafting a reply. Always with a fallback path defined, because a model that is having a bad afternoon should not stop an invoice being raised.

  • Data

    APIs and JSON

    The plumbing under all of it, and the reason your tools can be joined up at all.

    Almost every business system has a way in. Where one does not, there is usually an inbox, a spreadsheet or a file drop that will do. The first question on any build is which of those we are working with.

  • Data

    Dashboards

    So you can see what ran, what it did and what failed, without asking us.

    Every system we run reports somewhere you can look. Silence is what kills automations: the ones that quietly stopped six weeks ago are the ones nobody was watching.

  • Automation

    n8n

    Where most of what we build actually runs.

    Self-hosted or on their cloud, your choice. Every workflow exports as a file you own, which is why we use it: if you ever want to take a system in-house, there is nothing to unpick.

  • Automation

    Zapier

    For the small joins that are not worth engineering.

    Sometimes two apps need to talk once a day and that is the whole requirement. Building that properly would cost more than it saves, so we do not.

  • Automation

    Make

    When a client already runs on it and wants to keep going.

    We inherit plenty of half-finished Make scenarios. Usually the answer is to fix and extend what is there rather than rebuild it somewhere else on our preference.

  • Data

    Airtable

    The middle layer when a spreadsheet has outgrown itself.

    Half the businesses we meet run on one enormous spreadsheet nobody trusts. Airtable gives that data a shape an automation can rely on, without a database project.

  • Communication

    Slack

    So a system that stops tells somebody, in the place they already look.

    Every workflow we run reports into a channel: what ran, what it did, what failed. Silence is the thing that kills automations, and a channel is how you notice.

  • Engineering

    GitHub

    Every change recorded, and every version recoverable.

    Custom code, workflow exports and this website all live in version control. Nothing we build is a file on somebody's laptop, and every change has a date and a name on it.

  • Engineering

    Vercel

    Where the sites and the dashboards we build get served from.

    Fast by default, global, and it rebuilds itself when the repository changes. The site you are reading runs on it.

What that means for you

You own it

Every workflow exports as a file. Every line of custom code sits in a repository you can be given access to today. There is no proprietary layer of ours in the middle.

You can leave

If you ever want to take a system in-house or hand it to another agency, we export it, document it and help with the handover. That is written into how we build, not offered as a favour.

Nothing gets replaced to suit us

We work with whatever you already run far more often than we replace it. New software is a cost with a long tail, and preference is not a reason to make you pay it.

The usual questions

Do I need to buy these tools myself?

Usually not. Most of what we build runs on infrastructure we manage as part of the monthly fee. Where a licence has to sit in your name, we say so before we build rather than after.

What happens if I want to stop working with you?

Every workflow exports. Every repository can be transferred. We document the system and help with the handover, and we would rather do that well than have somebody stuck with us because leaving is hard.

Will you replace the software we already use?

Rarely. Replacing working software costs licences, training and two weeks where nobody can find anything. We connect what you have far more often than we swap it out, and we will tell you plainly on the few occasions replacing is genuinely the cheaper answer.

Something we use is not on this list. Can you still work with it?

Almost certainly. These seven cover most of what we build, but if a system has an API, or even just an inbox and a spreadsheet, there is usually a way in. Tell us what you run and we will say honestly whether it can be joined up.

Free automation audit

Don't know what to automate?

That is exactly what the audit is for. You know your business. We know what can be automated.

Get my free automation audit

Replies within one working day. No commitment, and a person reads it.

What we look for

Including the things that should stay exactly as they are.

  • Work that repeats
  • Bottlenecks where things sit and wait
  • Leads that go cold in a gap
  • Admin nobody should be doing
  • Where AI genuinely helps
  • Where it genuinely does not
  • Tools that should be talking and are not
  • What to leave exactly as it is

Not sure which of these your business needs?

Nor should you be. Tell us the job that keeps coming back and we will work out what it should run on.