How we work

Understand first. Price last.

We do not quote a number at a business we have not understood. Every build goes through the same four stages — and you can stop after any of them.

01Understand02Test03Deploy04Run
Workflow mapweek 01 · draft
Comes inWhat happens now
Phone call
Answered 9–5, if someone is freeafter 6pm → missed
WhatsApp
One person, on their own phonefine, while he is free
Website form
Email, then typed into a sheettyped twice
Receipt photo
Shoebox until month endthree weeks late
Three places work falls through≈ 6.5 hrs / week
I

Understand

One call, then a few days

First we learn the job. Nothing is sold yet.

We sit with how the work actually happens — who answers the phone, what gets typed twice, where things fall through. You get a written map of it back, whether or not you hire us.

  • A short call about the day-to-day, not the technology
  • We map the workflow and where the time goes
  • You get an honest read: automate, change the process, or leave it alone
Priced asFree
Pilot consolerun #14
Sandboxyour data · no customer is touched

Scenario 07 · WhatsApp

“Can you get to Galway tomorrow?”

“Friday covers Galway — order by Wednesday evening.”

intent: delivery.windowlanguage: en-IEwrote: nothing
  • Delivery window questionpass
  • Price and minimum orderpass
  • Caller asks for a humanpass
  • Accents and background noisepass
Scenarios passed27 / 30
II

Test

2–3 weeks

Then we prove it on your real workflow.

We build the smallest version that answers the question “does this actually work here?” — running on your real scenarios, with sandboxed data, so you can judge it rather than imagine it.

  • A working pilot, not a mockup or a slide
  • Your real cases, your wording, your customers' questions
  • You use it yourself before anything goes live
Cutoverv1.4 → v2.0
Phone lineWhatsAppWebsite chatStaff TelegramYour booksCustomer emailAssistantLIVE
Cutover 03:120 min downtime

old path kept warm · rollback one command away

III

Deploy

3–6 weeks

Then it goes live, carefully.

Once the pilot earns its place, we wire it into the real thing: your channels, your books, your staff escalation. We have done this on a live system without a minute of downtime.

  • Connected to the systems you already run
  • Escalation rules agreed with your team, not assumed
  • Dashboards, logging and retries from day one
Operationslast 30 days

Uptime

99.94%

Median answer

1.2s

Handled

412

Daily availabilityhealthy
  • Tue 14:03telegram.retryrecovered 2s
  • Wed 09:41prices.updated6 intents retuned
  • Fri 18:20new.workflowrefund requests
IV

Run

Ongoing

Then we keep it honest.

Businesses change — new products, new prices, new questions. We watch what the system gets wrong, fix it, and add the actions your team starts asking for.

  • Monitoring and incident response
  • Tuning as your catalogue and prices change
  • New workflows added as you find them
Priced asManaged

How we behave

We tell you when not to build

If the work is too rare or too varied to automate well, saying so costs us one call and saves you a project.

You see it before it is live

Nothing reaches your customers until you have used it yourself on your own scenarios.

Money is never guessed at

Anything touching an invoice, a payment or a customer is confirmed by a person first.

You own what we build

It runs on your accounts, with your data. If you walk away, it keeps working.

Tell us the job that eats your week.

We'll show you what it looks like automated — using your real workflow, not a generic demo. If it isn't worth building, we'll say so.