How we work

Map it. Scope it. Build it. Measure it.

We work on one process at a time. Before we build, we agree what success looks like. After launch, we look at what actually happened and improve from there.

Understand the process

We map what happens today, including exceptions and the systems involved.

How requests arrive, in what volume and when
Who handles them, and what gets typed where
The inboxes, property, freight and accounting systems in use
The exceptions that happen every week

Agree one useful starting point

We define the scope, success measures and what stays with a person.

One inbox and one workflow to begin with
A baseline, so there is something to compare against
What the workflow may draft and do, and what it may not
Setup and ongoing costs, with a usage alert or cap

Build and test together

It runs alongside your team on real examples before it touches a live system.

Built around your approved wording and rules
Run in shadow mode on the emails you actually receive
Reviewed by the people who will rely on it
Launched only when the checklist below is complete

Review what happens

We use real usage and feedback to improve the workflow.

A weekly review during the first month
Items handled, and staff minutes on each
Escalations, corrections and anything it got wrong
Time to first response, measured against the baseline

Where it starts

One shared inbox, such as maintenance requests or quote requests. The workflow drafts, and your team approves every output.

What comes later

Writing straight into your systems, once the drafts have proved accurate. Then the next workflow, such as chasers or invoice matching.

What you decide

The approved wording, the escalation rules, which contractors or rates it may use and who signs off.

How it ends

With a review against the baseline. You choose whether to roll it out, change it or stop.

Before anything goes live.

Every project clears the same checklist. We test the awkward cases, because those are the ones that matter.

Baseline and success criteria agreed
Common request types tested
Messy and incomplete emails tested
Out-of-scope requests tested
Wrong-match and low-confidence cases tested
Integration outages tested
Manual fallback in place
A simple way for you to pause it
Escalation rules and approved wording signed off
Data map and processing agreement signed
Usage alert or cap set with you

What stays with people.

Knowing where automation stops is part of the design. We write it down before we build.

Safety and legal matters

Gas, electrical and fire reports, legal notices and customs declarations follow fixed rules and go to a person. The workflow does not decide them.

Money and commitments

Quotes, spend and anything sent in your name wait for a person to approve them, and leave an audit trail.

Your customer data

Before real data is handled, we map where it goes and agree access and retention with you. We can support your data protection impact assessment.

Being clear about AI

Where your customers deal with an automated system, they are told. Messages can be approved by a person or kept to agreed templates.

Evidence of how it works

Rules, tests and logs are documented, so you can show an insurer, a regulator or a client exactly what the workflow does.

The off switch

Every workflow has a manual fallback and an easy way for you to pause it.