Notes on getting AI working in business: setup, cost control, and rollout.
Enter your headcount and a few assumptions to get annual seat cost, a usage allowance, a recommended hard spending cap, and the minutes per week each user needs to save to break even.
Read moreThe EU AI Act already requires businesses that use AI to ensure their staff have sufficient AI literacy. Who it covers, including some non-EU businesses, what the 2026 changes did to the timeline, and what proportionate training looks like.
Read moreThree different ideas of what a business agent is. Where Copilot Studio, ChatGPT and Claude each run, how each is billed, how much admins can control, and which kinds of job suit each one.
Read moreAustralian AI consultancies that publish their prices, the free AI Adopt Centres, what each suits, and the local points, from GST to the December 2026 privacy changes, to check before you hire.
Read moreFrom 10 December 2026, privacy policies must describe automated decisions that significantly affect people. Who is covered, why human review does not get you out of it, what to disclose, and a five step plan.
Read morePublished prices for AI consulting in 2026, grouped by engagement type: hourly advice, assessments, fixed projects, retainers and large programmes. Plus what each pricing model does to your risk.
Read moreNamed firms and options for getting AI implemented in a small or mid-sized business, ranked on criteria we publish, with each firm's own published prices and a clear note on where we sit in the list.
Read moreA budget for AI transformation built stage by stage, with planning figures for businesses of 30, 80 and 180 people, and the internal-time line that most budgets leave out.
Read moreSince April 2026 the two plans cost the same per standard seat. What actually separates them: seat mixing, spend controls, connectors, admin depth, and the kind of work your team does all day.
Read moreA side-by-side of the four business AI assistants as they stand in October 2026: what each seat costs, which admin controls sit on which tier, where spend caps exist, and the simple rule that decides most cases.
Read moreClients are adding AI clauses to contracts and insurers are adding AI questions to renewals. The common clauses, how to respond without losing the work, and how to write your own AI terms first.
Read moreSpreadsheet analysis, client letters, workpaper review and confidentiality obligations. How ChatGPT, Copilot and Claude compare for an accounting practice, and how to split licences by role.
Read moreYour staff already have Copilot Chat. What the paid licence actually adds, what it costs in 2026, and a three question test for deciding who should get one.
Read moreFor a small or mid-sized firm, the choice between Claude, ChatGPT and Copilot comes down to document work, confidentiality settings and where your files live. Plus where legal-specific tools are still worth paying for.
Read moreMost of AI transformation is work a capable internal owner can lead. The parts that are worth buying help for, the parts that are not, and how to keep any outside engagement small.
Read moreA copy-ready policy for AI agents and automations, covering ownership, access, approvals, spending limits, logging and incidents, with a note on how to adapt each section.
Read moreEvery large consultancy now sells AI transformation. When a big firm is worth the fee for a 50 to 500 person business, and when you are paying for a logo and a pyramid of juniors.
Read moreYour AI bill has jumped and nobody is sure why. The seven usual causes, where to look for each one in the usage data, and what to change.
Read moreA four quarter roadmap sized for a business of 20 to 200 people, with an owner, a measure and an exit test for each quarter, and the points where you should stop and reassess.
Read moreCopilot surfaces anything a user can already open. In most small businesses that includes things it should not. A one week plan to find the worst oversharing, fix it, and keep it fixed.
Read moreFifteen questions to put to any AI consultant, grouped by what they reveal, with the answers that should make you walk away. Use them on us too.
Read moreConnectors and MCP servers let AI read and act on your systems. A practical way to decide which ones to allow, how much access to give, and why read access and write access are different decisions.
Read moreStrip out the enterprise vocabulary and AI transformation in a smaller business comes down to three stages: everyone has a governed tool, a few workflows change, and the business is organised around what AI is now good at.
Read moreYour IT provider has offered to handle AI. Sometimes that is the best option. The parts MSPs do well, the parts they usually do not, and a clean way to split the work.
Read moreMost Workspace businesses are already paying for Gemini and have never configured it. When it is enough on its own, when a team should also get a standalone assistant, and what to switch on first.
Read moreSeats are a fixed cost. Agents are not. Why agent usage makes AI bills unpredictable, the four layers of spending control, and a sensible way to set each one.
Read moreFixed-price implementer, hourly consultant, your IT provider, a vendor partner, or yourselves. A straight comparison of the five routes, including the cases where hiring anyone is a waste of money.
Read moreFor a Microsoft 365 business the question is rarely Copilot or ChatGPT. It is which people need which layer. The three licence layers, the permissions problem nobody mentions, and a split that keeps the bill sensible.
Read moreOperations, finance, HR and customer service usually get AI licences last and benefit most. The tasks that pay, the process documentation win, and how to introduce it without frightening people.
Read moreHow to write client data rules for AI tools that people actually follow, the four categories that matter, the subprocessor question, and what to do if something goes into the wrong place.
Read moreWhy most internal prompt libraries are dead within a month, what to build instead, and how to write a task card that someone uses on a Tuesday without being reminded.
Read moreWho should pay for AI licences and usage, why chargeback usually suppresses adoption, and how to set departmental alerts that catch a problem without strangling the tool.
Read moreWhere AI genuinely helps a bid team, why responding to more RFPs is the wrong instinct, and the five stages where it saves real hours without flattening your win themes.
Read moreAdoption spikes then collapses around week five. What causes the relapse, the three numbers that predict it, and the interventions that actually pull usage back up.
Read moreWhy AI proposals stall in partnerships, the real objection behind each stated one, and how to write a one-page paper that gets approved rather than deferred to the next meeting.
Read moreHow professional services firms should handle client disclosure of AI use. Where it belongs, what to say, what never to promise, and how to answer a security questionnaire well.
Read moreWhat happens when you connect an AI tool to your document store, why permissions are the whole problem, and the order to connect things in so nothing surfaces that should not.
Read moreWhat actually differs between business and enterprise AI plans, which differences matter at 30 to 300 staff, and the four triggers that genuinely justify the more expensive tier.
Read moreA practical migration plan for consolidating staff onto one governed AI workspace. What transfers, what does not, how to run the amnesty, and how to stop people keeping the old account on the side.
Read moreHow advisory and wealth firms use AI without creating recordkeeping, supervision or marketing rule problems. What counts as a business record, and where the time actually comes back.
Read moreWhere AI helps a recruitment or staffing firm, why anything that screens or ranks candidates is regulated, and how to get the volume gains without inheriting discrimination exposure.
Read moreHow medical and dental practices use AI safely on administrative work. Why the business associate agreement is the whole question, what counts as PHI, and where the time actually comes back.
Read moreWhere AI helps a consulting firm, why the deliverable layer is the part at risk, what it does to your junior development pipeline, and how to make your own past work the advantage.
Read moreHow a real estate brokerage rolls out AI across an independent contractor agent base, the fair housing language risk in listing copy, and where the time actually comes back.
Read moreWhere AI helps a commercial insurance brokerage, how to keep it away from your errors and omissions exposure, and what to configure before account managers start using it.
Read moreHow marketing and creative agencies can use AI without producing the same work as everyone else. Client IP, contract clauses, where it helps, and where it quietly damages the product.
Read moreHow accounting and CPA firms can use AI on the work that actually compresses busy season, what it must not touch, and how to handle client data under the Safeguards Rule.
Read moreA practical guide to AI for law firms. Which tasks it genuinely speeds up, which it must never touch, how to handle client confidentiality, and what to configure before anyone starts.
Read moreAn AI agent takes actions, not just drafts text. What changes when software acts on your behalf, which jobs suit agents, the four controls they need, and when to wait.
Read moreThe nine mistakes behind most failed AI rollouts: buying before deciding, training on capabilities, no spending cap, personal accounts, and five more. With the fix for each.
Read moreA week-by-week AI rollout plan: what happens in the first 30 days to go live, what days 31 to 60 are for, and what to review at 90 days. With owners and exit criteria.
Read moreAn AI readiness assessment you can complete in an hour. Twenty-four questions across decision, data, technical, people and commercial readiness, with what each answer means.
Read moreHourly billing transfers estimation risk to the buyer. When fixed price is right for AI consulting, when it is not, and the questions that expose a weak proposal.
Read moreSmall businesses have advantages large ones do not. Eleven genuinely useful AI tasks for operations, quoting, admin and customer response, plus what to skip and what it costs.
Read moreAI for accounting, legal and consulting firms: the tasks that repay effort immediately, the billable hour problem, client confidentiality clauses, and what to do in the first…
Read moreWhen to configure an off-the-shelf AI tool and when to build something custom. Four questions that decide it, the real cost of building, and why most businesses should buy first.
Read moreTraining, retention and processing are three different questions, and most AI data worries confuse them. What to ask your provider, what to put in client answers, and what…
Read moreHours saved multiplied by hourly rate is a number nobody believes. Four honest ways to measure AI ROI, what to record before you start, and when to admit it is a cost of doing…
Read moreGeneric AI training produces nodding, not adoption. How to build role-specific task cards, run a 45 minute session that works, and fix the four reasons people quietly stop.
Read moreThe AI admin settings that actually matter: retention, training exclusion, sharing, workspaces, connectors and single sign-on. A configuration checklist you can work through.
Read moreRunning every task on a flagship model is the most expensive habit in business AI. How to build a model routing map, which jobs need reasoning, and what it saves.
Read moreSix structural reasons AI pilots never reach production, how to diagnose which one is holding yours, and the exit criteria that turn a pilot into a decision.
Read moreMost AI usage policies fail because they are written for lawyers, not staff. A one-page structure, the traffic light method, and the eight decisions a workable policy has to make.
Read moreShadow AI is staff using unapproved AI tools with company information. Why banning it fails, how to find out what is actually happening, and how to convert it into a governed…
Read moreAlerts tell you what you already spent. Caps decide the worst case in advance. How to set hard AI spending limits, pick the right number, and stop runaway AI bills.
Read moreA decision framework for choosing a business AI provider: the seven criteria that outlast model releases, why bake-offs rarely settle anything, and how to make the call in a week.
Read moreWhat AI actually costs a business: licence tiers, per-seat pricing, usage-based spend, setup fees and hidden costs, with worked examples for teams of 12, 50 and 200.
Read moreA practical guide to AI implementation for business: the five decisions that matter, a week-by-week 30-day plan, and how to know when the rollout is actually finished.
Read more