Practical notes on making systems work.
Tools you can use in a couple of minutes, and plain-language guides for business owners and operations managers, not for engineers.
Shadow AI Audit
Your team is already using AI. Twenty-four questions give you a risk score, your top five risks ranked by what would hurt the business, a 30-day fix list, and an AI policy you can circulate this week.
Start the audit →- A risk score out of 100, and a band
- Your top five risks, ranked
- A 30-day fix list
- An AI policy and use register
35 guides
Why one agent is the wrong shape
One assistant that can do everything is one account that can do everything. The case for splitting the work by job, with a worked example of what an agent team looks like.
Read →What an agent should not be allowed to do yet
Supervised is not a pilot you graduate from. How to decide, per automation, whether it has earned the right to run unattended.
Read →What an AI agent does all week
A concrete week: enquiries, customer questions, chasing payment, CRM insights, reporting and marketing drafts, with what stays with a person set out in a table.
Read →Chasing overdue invoices without chasing customers away
The accounts that go unchased longest are rarely the smallest. What to send at each stage, and where automation has to stop.
Read →What your CRM already knows
The questions people actually want answered are about absence, which is exactly what reports are bad at. Plus the first thing you will find, which is probably your data quality.
Read →Why AI marketing copy all sounds the same
The generic draft is the model accurately reflecting that you gave it nothing only you could know. What to feed it instead, and why a person always sends.
Read →The quote that lives in one person's head
A consistency problem and a key-person risk before it is a speed problem. What an agent can draft, what a person must still decide, and why the small jobs are about to become worth having.
Read →When AI changes what you sell, not just how you work
Efficiency or business model? One question separates them, and most businesses are sold the expensive answer when they needed the cheap one.
Read →How AI disrupts a business model: four patterns
Four patterns that cover most real business model disruption, the parts that tend to survive it, and three questions to locate your own exposure.
Read →Innovation and resisting change
Four reasons people resist change, only one of which is stubbornness, and the question that turns the argument from whether into how.
Read →What an AI agent actually is
Agent or automation, the six setup decisions that matter, and an honest account of what the time saving actually is.
Read →When the answer is not a bigger server
Why upgrading the server buys months rather than fixing anything, and where marketplace slowness actually comes from.
Read →How small business websites actually get hacked
The four ways in, why being small is not protection, and the patching job nobody in your business owns.
Read →Before you hand over your code
The four things you need to own, and what to ask a departing developer for before the relationship ends.
Read →The AI your team is already using
Your team is already using AI on personal accounts. How to find out what, and give them a sanctioned option instead.
Read →What Australian privacy law asks of you
Whether the Privacy Act applies to you, and the handful of principles an AI project actually touches.
Read →Why audit preparation takes a fortnight
Audit preparation as a capture problem, not a paperwork problem, and where AI belongs in care records.
Read →The gap between finishing a job and invoicing it
Why invoicing waits on paperwork, and what capturing work on site actually has to do to get used.
Read →When the paperwork arrives faster than you can file it
Automated document intake for claims work, and why a system should be judged on what it refuses to guess at.
Read →Everything written down twice
Why re-keying costs more than the typing, and how to decide which system holds the truth about a client.
Read →One view of the job
Variations, site capture and job visibility, in the order worth fixing them for a building or trade business.
Read →Before you build a dashboard
Power BI, or something lighter? Exploring your numbers and simply seeing them are different jobs, modelled differently underneath.
Read →Cloud AI, or your own server?
Running AI in the cloud versus on hardware you own: what each gives you, what it really costs, and when in-house genuinely wins.
Read →Before you automate anything
Four questions to ask about a process before you spend money automating it.
Read →Spreadsheets are not the problem
When a spreadsheet is fine, and the three signs it has become a liability.
Read →What AI actually costs an SME
A plain breakdown of setup, running and support costs for a first AI project.
Read →Keeping your data where it belongs
The questions to ask any vendor about where your business data is processed.
Read →Where AI should, and should not, make decisions
How to decide where AI can act alone and where a person must stay in the loop.
Read →Choosing your first automation
How to pick a high-volume, rules-based process to automate first.
Read →Why we fix the process before the software
Why sorting out the process first makes the software cheaper, or unnecessary.
Read →Getting your systems to talk to each other
The real cost of disconnected tools, and where to start joining them up.
Read →What to ask before replacing a legacy system
Modernise or replace? The questions to answer before you commit, and why staged beats all-at-once.
Read →Keeping AI working after launch
Why AI drifts after launch, and what monitoring and ownership it needs to keep working.
Read →A buyer's guide to custom business tools
Build versus buy, and how to tell when off-the-shelf software has stopped fitting.
Read →Claude, OpenAI and DeepSeek explained
What the differences between the main AI platforms mean for your business, in plain terms.
Read →Get new guides as they are written.
Plain notes on AI, automation and making systems work, for people running a business rather than a development team. The same thing you have just been reading, sent when there is a new one.