Learn before you spend.
Guides written for decision makers — plain English, no hype, no gate where it doesn’t earn it. This centre grows monthly; three foundation guides are below in full.
Read the guides
Plain English, no hype, no gate — three foundation guides in full below.
01
Put your own numbers on it
The ROI calculator computes from your inputs — nothing asserted..
02
Talk it through, free
Thirty minutes of honest advice — even if the answer is “not yet”.
03
A plain-English guide to AI automation
What it is. AI automation means software doing the repetitive, rule-shaped parts of office work: reading a document and posting its contents into a system, routing a request to the right person, compiling the weekly report, answering the question that has been answered a thousand times before. The “AI” part matters where the work is messy — documents that don’t follow a template, questions phrased a hundred ways — because modern models handle that messiness where older automation couldn’t.
What it is not. It is not a robot workforce, it does not require replacing your systems, and it is not intelligence you can leave unsupervised. Well-designed automation keeps people in the loop exactly where their judgement matters: reviewing the extracted invoice before it posts, approving the drafted letter before it sends.
How to tell whether you have a use for it. Ask three questions of any task: Does it repeat weekly or daily? Does it follow rules a sensible new starter could learn? Does it move information between systems or formats? Three yeses is a candidate. Then ask the only question that matters commercially: what does this task cost in salaried hours a year? If that number would embarrass you in a board meeting, it is worth a proper assessment.
How to build the business case for automation
The arithmetic a finance director will accept. Start with hours: how many people touch the task, for how long, how many weeks a year. Multiply by a loaded hourly cost (salary plus overheads — annual cost divided by roughly 1,600 working hours is a fair proxy). That is the visible cost. Add the two invisible ones honestly: error rework (manual steps create mistakes that cost twice — once to make, once to fix) and decision delay (what it costs when the numbers arrive after the meeting).
The claims to avoid. Do not import percentages from vendor case studies — your finance director has seen them too and discounts them on sight. Do not claim savings from redundancies you will never make; automation returns hours, and hours only become money when they are redeployed to work that earns. And label every estimate an estimate: credibility survives a cautious forecast, not an inflated one.
The structure that gets signed off. One page per opportunity: the process, its current annual cost (your own figures), the proposed change, implementation cost, payback period, the risk and its mitigation — and, crucially, the measurement plan: which number will be tracked, from when, by whom. A business case with a measurement plan says “hold us to this”, and that is what boards fund.
Choosing your first process to automate
Why the busiest bottleneck usually isn’t the right first pick. The most painful process is usually painful because it is complex, political, or tangled across departments — the worst possible place to learn. First projects carry a second job: proving to your own organisation that this works. Pick a process where success is likely and visible, not merely valuable.
What to look for instead. High volume and high repetition; clear rules with few exceptions; one owner who wants it fixed; data already in systems rather than on paper; and a result you can measure within weeks — hours returned, errors reduced, days-to-bill shortened. Invoice handling, enquiry triage and report compilation are classic first wins precisely because they score well on all five.
How to run it. Agree the success metric before anyone builds anything. Pilot on live work with a human reviewing every output until the error rate earns autonomy. Measure against the forecast, publish the result internally — including what disappointed — and let the evidence, not enthusiasm, decide what gets automated second.
More guides, checklists and a glossary are published here as they are written — one useful email a month announces them (sign up in the footer). No gated PDFs where a web page does the job.
Read enough? Put a number on it.
The ROI calculator takes five minutes and asserts nothing it can’t compute from your own inputs.