AI set-up and implementation
A new service, built on the system we run our own firm on: the right AI tools, workflows for the repetitive tasks that fill your week, written rules for what your data may and may not touch, and training so your people stay in charge.
In plain language
An AI assistant answers when you ask, and once it is connected to your email and files it can also act, which is why its settings matter. An AI agent is given a goal and the tools to reach it: it reads what it needs, works out the steps, uses your software to do them, and reports back.
That makes agents good at work that breaks into small, well-defined steps: sorting an inbox, filing documents, turning meeting notes into tasks, preparing a first draft. It also means an agent can do the wrong thing confidently, which is why every set-up we build keeps a person approving anything that leaves the business.
1
Reads the email, document, calendar or record it has been given access to.
2
Works out the steps to reach the goal, using the instructions you set.
3
Drafts, files, updates or prepares, inside the tools you already use.
4
In every set-up we build, a person approves before anything is sent, paid or submitted. Agents do not stop for a person on their own, so the check is built in.
Start here
A look at where your hours go, which tasks suit AI and which do not, what data is involved, and which tools keep it in Canada. Where a question turns on privacy law, we say so and you take it to a lawyer. You get a written plan, quoted before work starts.
Tools
Business accounts for tools such as Claude, ChatGPT or Microsoft Copilot, connected to Google Workspace or Microsoft 365, with the privacy settings checked rather than left on default.
Workflows
Inbox sorting and reply drafts, document intake and filing, meeting notes into task lists, deadline tracking, research summaries, and bookkeeping preparation from your Xero data, with every entry posted by a person.
Rules
What data may go into which tool, what is removed first, who approves outputs, and how long anything is kept, written for your business as a working policy. It is not legal advice: what privacy law requires of you is a question for a lawyer.
People
Hands-on sessions built around your own tasks: how to brief an assistant, how to check its work, and when not to use it.
After
Adjusting the workflows once real work has run through them, and a review when the tools change, which they do often.
Our own first client
Before we offered this to anyone, we built the core of it for ourselves.
No AI set-up we build sends an email, a payment, a filing or a funding application on its own. It drafts; a person decides.
When an AI tool analyses your financial statements, names, addresses and banking details are taken out first. Agents that work inside your inbox, files or Xero see that information as they work, so we connect them only through business accounts whose privacy settings we have checked, and we tell you what each one can see.
Every core figure an AI surfaces from financial records, such as revenue, net income or margins, is checked by hand against the original documents before it reaches a client.
Anything AI drafts for a client or for publication that states a fact is checked separately before it goes out, by a step whose only job is to assume it is wrong and find out.
We tell you which tools store or process information outside Canada, so the choice is yours and your privacy notice can say so.
AI does not make tax, legal or funding decisions for you, and we do not set it up to.
Start here
Tell us where the hours go. You leave with a first view of which tasks might suit AI, and what a readiness review would involve.
