Build logs and guides from real Kove projects. What works, what it costs, and what to avoid. Written from the work, not theory.
Comparing two specific tools? Every comparison is indexed here, with the one thing that decides each.
Start with what the market actually offers, then work out whether building beats buying.
The admin lane: where the hours go, and which of them a system can take.
Licensing first, then the data layer, then the agent stuff everyone is asking about.
Reference pages that sit outside the post list: the migration hub, Power Platform pricing in NZ, Dataverse pricing, end-of-support dates, the migration checklist and the trades software price index.
Power BI has no built-in Xero connector. The three ways a small NZ firm gets Xero data into a report, what each costs, and which to pick.
What a 5 to 20 person NZ law firm builds on Microsoft 365 (matter intake, the conflicts and compliance register, AML document collection), what practice software already does, and the licence line.
The three builds that pay for a 5 to 20 person practice (client onboarding, document chasing, the compliance calendar), the licence line, and what Xero already does.
The three systems worth building on the Microsoft 365 a 5 to 30 person engineering consultancy already pays for, what each costs, and what to leave in Excel.
Power BI Pro is NZ.60 per user per month. Who actually needs it, how it reads Dataverse and SharePoint, and the three reports to build first.
Azure SQL is the honest alternative when the logic lives in T-SQL. What each costs a 5 to 20 staff NZ firm, and the one row that decides it.
Free updates end 12 January 2027; three paid years announced, no price yet. What a 5 to 20 staff NZ firm does with the box, and when to skip the ESU.
Express stops accepting writes at 10 GB per database. What breaks, how to buy time, and when Microsoft 365 is the cheaper exit for a small NZ firm.
Support ended 14 October 2025. The .accdb file is fine; the program opening it is not. Three ways forward for a small NZ firm, priced.
USD 10 per person per app per month, only in months they open it, billed through Azure. The maths against NZ.40 Premium, and the break-even.
NZ.40 per user per month plus an implementation. Most firms under 20 staff asking about it need a job system on Xero, not new accounting. How to tell.
When a Microsoft List is enough, when a Form and a flow beat both, and when a Power App earns its build cost. All on the licence you already pay for.
If you already pay for Microsoft 365, Power Apps on SharePoint costs nothing extra. When the others still win, and the yearly cost for 10 people.
NZ.60 a month per site for 100 logged-in users, NZ.30 for 500 anonymous. What that buys, the cheaper alternatives, and when to build one.
How a photo taken on site becomes the proof on the invoice without anyone renaming files: the tools that do it, and the one habit that makes it work.
The signs a spreadsheet has quietly become your system of record, the risks that come with it, and the order to move things out of it.
The arithmetic inside the Kove ROI calculator, why 90 days is the bar, and how to run the test on your own numbers before you build.
How each pricing model shifts risk between you and the builder, what fixed scope actually means, and why I price the way I do.
Twelve questions that separate a builder from a reseller, what a good answer sounds like, and the ones I would want to be asked.
Five things ChatGPT genuinely does well for a NZ trade business, three ways it goes wrong, and the one rule that keeps client data out of it.
What an AI phone answerer can and cannot do for a NZ trade business in 2026, what it costs, and the missed-call maths behind the decision.
Plain-English arithmetic for a small NZ business: what a token is, what a real agent burns in a day, and the monthly bill in NZD.
Bank rules, repeating invoices and invoice reminders are already in your Xero plan. What each one does, how to set it up, and the hours it hands back.
What Hubdoc and Dext actually do with a supplier invoice, what each costs a NZ business, and when typing it in is still the right answer.
The Xero add-ons worth paying for in a NZ trade business, only tools already checked on this site: job management, receipt capture, and what you already have.
Xero Projects is the obvious landing spot after WorkflowMax. Where it fits a NZ services firm, where it falls short, and what to do about the gap.
LBP records, consent inspections, variations and progress claims: what NZ builders need from job software.
Gas certificates, PGDB licensing and backflow records: what plumbers and gasfitters need from job software.
CoCs, ESCs and call-out volume: what NZ electricians should judge job software on.
Four coding agents, one person, and the rules that stop them making a mess.
A standard plug socket between an AI assistant and your systems. What it is, and what it is not.
Store your process in Dataverse so an agent follows it instead of guessing.
Eight task skills for a coding agent, from schema changes to bulk data loads. What each one covers.
How to switch on the Dataverse MCP server, what an agent can reach, and who pays for the calls.
Up to 10 makers in one canvas app at once. What it gives you, and the ten things it turns off.
A kitchen quote is six jobs pretending to be one. Where the time goes for a small NZ joinery shop, and which steps automate cleanly.
What the Tradify and Xero integration moves in each direction, what it never touches, and where a small automation fills the gaps.
When the Excel file that runs the business should become a SharePoint list, a Power App on Dataverse, or stay a spreadsheet. With NZD costs.
A 7-person firm does not need an enterprise agreement to move a database into Microsoft 365. What the large partners are built for, and the alternative.
Free with Microsoft 365 until it is not. The NZD prices, the connectors that trigger them, and worked totals for 5, 10 and 20 people.
Support ended 14 July 2026. What that means for a small NZ business, what paid updates cost, and the three realistic ways off it.
What Microsoft's Access migration tool handles, what it does not, whether SharePoint or Dataverse is the right home, and what it costs a small NZ firm.
The last security updates end 13 October 2026. The four options for a 5 to 20 staff NZ firm still running one, and what each costs.
Which parts of a SQL Server system move to SharePoint or Dataverse, what has to be rebuilt, and what it costs a 5 to 20 person NZ firm.
MCP server GA, a Python SDK, agent identities in Entra, and business data inside M365 Copilot. The 2026 Dataverse changes that matter, minus the launch-video gloss.
Excel answers what happened to one job. Power BI answers which jobs keep making you money. The crossover, and the licensing that catches firms out.
One is a per-seat chat assistant, the other a consumption-billed agent platform. Different tools, different bills, and the test for which one your problem needs.
The Approvals app covers one-step sign-offs without building anything. The moment you need stages, conditions or audit depth, you need a flow. The line between.
Dataverse for Teams is free with M365 and capped hard: 2GB a team, about a million rows, no API. Where those walls sit and when to upgrade.
Pulled from general sale in January 2026, kept for EA renewals, quietly restored for CSP in April. The full timeline and what to do if you hold it.
GeoOp and GeoNext stopped taking new users after the Access Group takeover. What that means for NZ crews still on it, and how to pick a landing spot.
Tradify is priced and built for jobs. SimPRO is quoted and built for projects. The crossover is the shape of your work, not just the size of your crew.
SimPRO does not publish prices. What a quote is actually built from, what NZ reviewers report paying, and the questions that keep a quote honest.
What Tradify, Fergus, ServiceM8 and NextMinute each get right about quoting, which tier the good quoting features sit in, and when none of them is the fix.
NextMinute bills by crew band: $199 a month covers three to nine people. Why band pricing beats per-user for a growing crew, and where the band jumps land.
Fergus starts around $53 a month, Professional from $75, timesheet-only seats from $26. How the plans split, and when the base plan is genuinely enough.
Tradify runs NZ$48, $52 or $62 a user a month ex GST across three tiers. What each tier adds, the add-on costs, and where per-user billing stops making sense.
Three price levels: features you already pay for, connector glue, and a built-for-you flow. Which level your problem sits at, and the maths that decides it.
$0 if a connector already exists, tens a month for glue, $4,000 to $8,000 built custom. What the money actually buys, and when not to spend it.
Roughly NZD 35 to 52 a user a month. The seat-by-seat test that shows who clears the bar, and the cheaper tools people usually mean when they ask.
M365 Copilot is a per-seat assistant over Office. A Copilot Studio agent is a purpose-built bot for one job, priced by usage. Mixing them up means buying the wrong one.
Copilot Studio is priced by consumption, not seats. A worked NZ cost estimate, what counts as a billed credit, and when a Power Automate flow is cheaper.
Not every automation needs a chat window. Three jobs that look like Copilot Studio agent work but run cheaper and more reliably as a Power Automate flow.
How Copilot Studio checks permissions when an agent is grounded on SharePoint, where oversharing sneaks in, how to scope it properly, and how to test it safely.
Why a Copilot Studio agent gave confident wrong answers, and how I fixed retrieval by restructuring the source content instead of rewriting the prompt.
How Copilot Studio message billing works: what counts as a message, where capacity disappears fastest, and how to cap and monitor cost before it surprises you.
Most firms under 50 staff default to Dataverse because it sounds more proper, then pay for capability they never use. One question actually decides it.
Canvas, model-driven and code apps solve the same problem three ways. Pick on the shape of your data rather than on which one demos best.
Power Apps code apps ship a real React front end with Dataverse and connectors behind it, GA since Feb 2026. What you keep, what you take on, and when canvas still wins.
Power Apps quotes for the same brief vary wildly in NZ, because the build is rarely the biggest line. Licensing, connectors and Dataverse usually are.
A flow that works fine in testing can suddenly need a paid licence for everyone who touches it. What's premium in Power Automate, and how to check first.
Power Automate, Zapier and Make all move data between apps. For most businesses the right pick comes down to what you already pay for, not which tool wins.
Power Platform is built for internal workflow, not high transaction volume, public apps for anonymous users, or jobs an off-the-shelf tool already does.
Business Basic, Standard or Premium for a 20-person firm: what changes, which security features earn their keep, and what Power Platform already includes.
Every Power Platform tenant accumulates flows no one remembers building. What breaks when the owner leaves, and how to inventory a tenant for free.
The 5,000-item list view threshold isn't the real limit; an index fixes that. Lookups, concurrency, delegation and reporting are what actually stop a list.
Dataverse security roles explained through one real scenario: a consultant, a team lead and a practice manager who each need to see something different.
Power Platform ALM explained at solo-builder scale: which environments you need, managed vs unmanaged solutions, and Pipelines instead of manual exports.
Xero, Tradify, Fergus and ServiceM8 all almost do job costing. What each one gives you, the timing gap that breaks it on staged jobs, and when to fix it.
Trade tools nail invoice sync to Xero. Almost none automate the quote itself, the slowest, most margin-critical step in winning the job. The few exceptions, and what they cover.
ServiceM8 charges by job volume, not per seat. Here's exactly where that crosses over per-user pricing like Tradify, worked in real numbers. NZD, July 2026.
NextMinute suits staged residential builds with progress claims. Tradify suits fast turnaround jobs. How to tell which one fits you, plus 2026 NZ pricing.
Ascora and NextMinute both handle staged building projects with variations and progress claims. How they differ, real NZ pricing, and the support gap on one of them.
A Workflow Review is a paid, two-week engagement that ends in a written plan: what to automate first, what it saves, what it costs. What the two weeks look like.
Command mode, not dictation, is the interesting part of my voice app: a local model turns speech into an action, entirely on-device, nothing sent to the cloud.
Quoting, scheduling, timesheets, Xero sync and pricing at 5 to 20 staff, worked out with real numbers. A straight read from someone who sells neither.
Tradify, Fergus, ServiceM8 and NextMinute compared by someone who sells none of them, plus the option nobody puts on the list: building the tool around how you work.
Ease of use against job-costing depth, pricing, Xero sync, and where ServiceM8 fits instead. I don't resell either one.
The measure, quote, PDF, retype loop that eats your evenings. What off-the-shelf tools fix for free, and where a custom bridge earns its keep.
The honest maths: weighing a build against subscriptions, why the cheapest 48-hour app rarely is, and when the numbers work.
A local builder who takes the admin off Canterbury trade businesses by fixing the gaps your existing tools leave, not selling you another subscription.
A self-updating Copilot Studio agent on a large internal Power App. The pattern, the 85% time saving, and the part most people get wrong: keeping an agent's knowledge live.
Two real, anonymised Copilot Studio use cases: turning a one-person spreadsheet into a shared knowledge agent, and automating the monthly policy-change check. Plus where an agent doesn't make sense.
Two years after Xero retired WorkflowMax, plenty of NZ firms are unhappy where they landed. Your three real options, the data problem nobody mentions, and when a custom tool beats another monthly invoice.
An overnight pipeline across 14 NZ directories. 290 unit tests, ~$8 in API credits.
A local voice layer: dictate into any app in under half a second, and hear anything read back in a clone of your own voice.
Book a free 30-minute call. I'll tell you what I'd automate first, and you'll get a straight answer either way.
