AI Operations

After launch day, someone is still watching.

Agents and automations work around the clock, and like any part of a business they need someone looking after them. We watch every system we built for you, fix what breaks, re-tune what it costs to run, ship an improvement each month and set it all down in a written report.

Everything in writing. A written reply within one business day.
ops / care-log running
DR
Daniel R. — three systems under care
support agent · order flow · voice line · sample data
watching
every step written to the audit trail step 0 / 6
One month of care

A month on the systems you run, from the first nightly check to the report.

Interfaces change, providers reprice, volumes grow and customers find cases nobody planned for. A sample month for Daniel R., whose support agent, order flow and voice line are in our care.

WEEK 1iHayz

Nightly checks on every system

Each system named in the proposal is checked every night: did it run, did it answer, did anything fail. A failure alerts us before it reaches a customer.

health · nightly
WEEK 2iHayz

A change upstream, caught and patched before it failed

A messaging service changed one of its template rules. The order flow was patched that morning, before a single run failed.

patch · before failure
WEEK 3Daniel R.

An improvement shipped and approved

The returns reply lost two steps. It ran on staging first and went live once Daniel approved it.

improvement · staging first
DAY 1report

The written report, on the first of the month

What ran, what broke, what improved and what comes next, set out in one document you can keep.

report.pdf
What looking after means

Four jobs, done every month.

Monitoring, fixes and a monthly report in writing are part of every engagement, along with the cost watch and the monthly improvement.

Monitored daily

Health checks run on every system each day. When something fails, the alert reaches us first, before it costs you an order or a customer.

Costs watched

When a provider changes its prices, the system is re-tuned so the running cost stays low for the volume you have.

Improvements, not only fixes

Every month one change ships, so a system runs faster, costs less to run or handles more volume.

Reported in writing

What ran, what broke, what improved and what is next, in one written report at the start of each month.

Where it stops

Four things we never change on our own.

Looking after a system means working inside limits you set. These stay with you.

01

What the agent says to customers

A change to the wording your customers read waits for your approval first, as the refund replies did in the sample week.

02

Money rules

Refund limits, payment steps and approval rules are never changed without a named person on your side agreeing.

03

Anything not listed by name

The systems we look after are named in the proposal. Anything outside that list stays outside until you add it in writing.

04

Your accounts

They stay in your name, and we work inside them. The accounts, the data and the documentation stay yours whether or not we look after them.

A month of operations

Watch, fix, improve, report, and back to watching.

The loop runs around the systems named in your proposal. Any change your customers would notice stops for your approval before it goes live.

A monthly loop around your systems, named in the proposal. Watch: every system, every night. Fix: before it costs you. Improve: one change a month. Report: in writing, monthly. Beside Improve, your approval comes before a change goes live.
How the engagement is shaped

Sized to the systems you run and how fast you need an answer.

The level is set in the proposal from the systems in scope and the response you need. Each one carries the monitoring, the fixes and the monthly report.

One system, kept running

Monitoring, fixes, a watch on running costs and the monthly report for a single system, with a response by the next business day.

Several systems, improving monthly

A set of systems stated in the proposal, one improvement shipped each month, a place in the priority queue and a same-day response when something fails.

Every system, with a roadmap

Dedicated capacity for everything we built, a roadmap, a same-day response, a quarterly review of where the systems go next and a direct line to the person who built them.

Stopping

There is no minimum term. One month’s notice in writing ends it, and the accounts, the data and the documentation stay yours.

Every engagement

Four things you always get.

Whatever the size of the work, these come with it.

01

A written brief

What we agreed to build, in one document.

02

A fixed price

Set in the proposal, before work starts.

03

Staging + approval

You see it and approve it before it goes live.

04

Documentation

How it works and how to change it, handed over.

FAQ

Before we take over the watch.

Send any other question in writing; a written answer follows within one business day.

What counts as one system?

An agent, an assistant, a voice line or a workflow: anything with one defined job, as scoped in the proposal that built it. A support agent and the order flow it passes work to count as two.

Is there a minimum term?

No. The engagement runs month to month and ends with one month’s notice in writing. The accounts, the data and the documentation are yours throughout, and stay with you after it ends.

Do you look after systems someone else built?

Yes, after a paid review. We read how the system is built and how it runs, then send either a quote to look after it or a written recommendation to rebuild it, with the reasons.

How fast do you respond?

By the next business day when we look after one system, the same day with priority for several, and the same day from dedicated capacity when we look after all of them. Most failures are caught and fixed by our monitoring before you notice anything.

Start

Not sure what you need? Ask in writing.

Describe the work in a few lines. We will reply in writing within one business day with what we would build and how.