Term not here, or still not clear?
Ask on a free 30-minute call. No prep needed, and no obligation to book anything after.
Microsoft Power Platform jargon and trades job-management terms, in plain English. No jargon used to explain jargon.
Microsoft's tool for building a business app without starting from a blank codebase. There are three ways to build one: a canvas app, a model-driven app, or a code app. Which one fits depends on your data more than which looks nicest in a demo.
Microsoft's tool for moving data and triggering actions between systems automatically, for example creating a job in your CRM and emailing the customer the moment a form is submitted. It plays the same role as Zapier or Make, and for a Microsoft 365 shop it's usually already paid for.
Microsoft's structured database behind Power Apps, Copilot Studio and, increasingly, Power BI, with permissions and relationships built in rather than bolted on. A lot of firms default to it because it sounds more proper than a SharePoint list, then pay for capability they never touch.
Microsoft's tool for building a chat agent that answers questions or does a task using your own documents and data, not a general internet search. It's a different product from the Copilot button already inside Word and Outlook, which is Microsoft 365 Copilot.
The unit Copilot Studio bills by. Every message or automated action an agent takes can use a credit, and the rate depends on how the agent was built, so cost isn't a flat monthly number the way a licence is. Most surprise bills come from an agent burning credits on a task that would run cheaper as a plain Power Automate flow.
The AI assistant built into Word, Excel, Outlook and Teams, licensed per person on top of an existing Microsoft 365 subscription. Not the same thing as Copilot Studio, which is for building a custom agent rather than using the one already in your apps.
A Power Apps build where you place every screen, button and data connection yourself, closer to designing a page than filling in a template. Full control over layout, at the cost of more of it to maintain; suits an app that doesn't map cleanly onto one clean set of data. More on picking by data shape.
A Power Apps build where the screens generate themselves from your Dataverse tables, relationships and business rules, rather than being laid out by hand. Faster to build and easier to keep consistent when the data behind it is well-structured, less flexible when you want a specific look.
A newer option, generally available since February 2026, that ships a real React front end backed by Dataverse and Power Platform connectors instead of the low-code canvas editor. Suits a team that wants proper source control and a custom interface without giving up the platform's data and security layer underneath.
A legacy system is the software a business still depends on but can no longer easily change: the vendor stopped updating it, the person who built it left, or the platform underneath (an old SQL Server, an Access file, a Windows Server 2012 box) is out of support. Legacy describes risk, not quality, and the fix is usually a migration, not a rewrite.
Software and data running on a server your business owns, in your building, rather than in a vendor's cloud. Microsoft 365 is the opposite: the servers are Microsoft's. The on-premises data gateway is the bridge, letting Power Apps and Power Automate reach a SQL Server on your own network without opening inbound ports, at the cost of a premium licence per user.
Moving a system to new infrastructure without changing how it works, for example copying a whole Windows Server into an Azure virtual machine. Fast, and it changes nothing, which is both the appeal and the catch: you still patch, back up and pay for the same system, just somewhere else. The alternative is migrating the data and rebuilding the front end on the new platform.
The software that runs one specific part of a business: the job register, the quoting tool, the stock system. Often bespoke, often built years ago on Access or SQL Server, and often the one thing everyone uses that nobody supports. In Microsoft 365 the usual replacement is a Power App on SharePoint lists or Dataverse.
A column, or set of columns, that Dataverse treats as unique besides its own row ID, such as a job number. Migrations depend on it: a dataflow matches child rows to parents through the alternate key, and unique indexes from SQL Server or Access have to be recreated as one by hand, as the data-type map notes.
A Dataverse column that generates a sequential value for each new row, with a prefix and a seed you set, so a job or invoice sequence can continue from where the old system left off. It replaces a SQL Server identity column or an Access AutoNumber during a migration.
A column that links a row to a row in another table. In SharePoint a lookup is a pointer with nothing enforced; in Dataverse it is a real relationship the platform enforces, which is why SQL foreign keys become Dataverse lookups and why orphaned rows are possible in SharePoint but not in Dataverse.
A scheduled or on-demand import that reads a source (SQL Server, Excel, SharePoint, a web API) through Power Query, cleans it, and loads it into Dataverse tables. The standard tool for moving a SQL Server database into Dataverse, through the gateway if the server is yours.
A Windows service installed on a machine inside your network that lets Power Apps, Power Automate, dataflows and Power BI reach data on your own servers, such as a SQL Server database, using outbound connections only. It needs a premium licence on the Power Platform side.
The storage pool a tenant gets with its Power Platform licences: 10 GB of database, 20 GB of file and 2 GB of log with the first subscription, plus 250 MB database and 2 GB file per Premium licence. Extra database capacity is NZ$64.70 per GB per month; the full picture is on Dataverse pricing in NZ.
A Dataverse column whose value is calculated from other columns using Power Fx, the way a spreadsheet cell holds a formula. It replaces SQL computed columns and Access calculated fields after a migration.
The formula language of the Power Platform: the Excel-like expressions used in canvas apps, formula columns and increasingly in flows. If you can write a spreadsheet formula, you can read Power Fx.
The option to bill Power Apps, Power Automate and Power Pages use to an Azure subscription by the month instead of buying licences: USD 10 per active user per app, USD 0.60 per premium flow run. For a single app it usually beats Power Apps Premium; from two apps the licence wins, worked in the pay-as-you-go post.
Microsoft's tool for building a public or client-facing website on top of Dataverse: a client portal, a supplier portal, a booking form that creates records. A separate subscription per site, priced in what Power Pages costs a small NZ business, and not included in Microsoft 365.
The free tier of Dataverse included with Microsoft 365 licences that have Teams: 2 GB and about a million rows per team, apps and flows that live inside Teams, fewer column types and no premium connectors. The limits that bite before the upgrade.
The daily cap on Power Platform actions per person or per flow: 40,000 a day for a Premium user, 6,000 for a Microsoft 365-only user, 250,000 for a flow on a Process licence. A flow that loops over a large list the wrong way can hit it by lunchtime.
Microsoft's paid, last-resort security patches for a product after its support has ended, sold in yearly steps for up to three years, Critical fixes only. Every date is on the end-of-support dates page.
The front end for SharePoint lists: a table with typed columns, views, rules, alerts and a built-in form, included in Microsoft 365. The first thing to try before building an app; Lists, a Power App, or a Form sorts which you need.
One of the Power Platform connectors included with Microsoft 365: SharePoint, Outlook, Teams, OneDrive, Forms, Planner, Excel Online and many more. A flow or app that uses only standard connectors costs nothing beyond the Microsoft 365 seat; the moment it touches a premium connector it does.
A Power Automate automation that runs in Microsoft's cloud: automated (fires on an event), scheduled (on a timer) or instant (a button or an app press). What most people mean by a flow, as opposed to a desktop flow that drives a Windows PC.
Power Automate's robotic process automation: a recorded or scripted sequence that drives a Windows program by clicking and typing, for software that has no API. Attended desktop flows come with Power Automate Premium; unattended ones need a Process licence.
Validation or logic set on a Dataverse table without code: require a field when another has a value, show or hide a field, set a default, block a save. It replaces the simpler stored procedures, triggers and Access macros after a migration.
Microsoft's reporting tool: dashboards and reports over Dataverse, SharePoint, SQL Server, Xero and most other sources. Licensed per user separately from Power Apps; the usual replacement for SQL views and Access reports after a migration, and the subject of Power BI vs Excel.
Where a tenant's environments, capacity, licences, pay-as-you-go billing and data policies are managed. For a small firm it is a page the owner opens twice a year, and the place to check Dataverse capacity before buying any.
The package Power Platform uses to move an app, its flows and its data structure between environments instead of rebuilding it by hand each time. Unmanaged solutions are still editable and used for development; managed solutions are locked, and that's what gets imported into test and production.
A separate, walled-off space for Power Platform apps, data and flows, typically one for development, one for testing and one for production, so a change doesn't reach real users before it's been checked. Solo builders often skip this and build straight in production, which is fine until the day it isn't.
The discipline of moving a Power Platform build from a developer's environment through test into production in a way that's repeatable and can be rolled back, using solutions and pipelines rather than copying changes by hand. Matters even for a team of one.
Whether a query against a large SharePoint list or Dataverse table gets processed on the server, which is fast and complete, or pulled down locally into Power Apps first, which is slow, capped, and often quietly wrong past a couple of thousand rows. A delegation warning in Power Apps is the most common reason a canvas app returns an incomplete result without telling anyone.
The set of permissions in Dataverse that decides what a given person can see, create or change, for example whether they can view only their own records, their team's, or everyone's. Getting a role wrong is the usual reason a shared Power App shows two people different amounts of the same data.
A Dataverse structure for grouping people, by team, region or division, so a security role can be applied to the whole group instead of one person at a time. Only worth setting up once an organisation has more than one team that genuinely needs different access, not by default.
Runs the day-to-day of a trade or field service business: quoting, scheduling, job tracking and invoicing, usually in one place instead of a paper diary and a spreadsheet. Tradify, Fergus, ServiceM8 and NextMinute cover most of the NZ market; none of them fit every business exactly.
Tracking what a specific job actually cost, materials, labour and overhead, against what it was quoted, so a business can see which jobs make money and which don't. Most job management tools claim to do this; fewer actually reconcile it against real invoices and timesheets without manual cleanup.
Going back over a job's actual costs against the original quote after it's finished, usually to check the margin or to price the next similar job more accurately. It only works if it happens close to the job finishing; done weeks later from a spreadsheet, the numbers are already stale.
An invoice for part of a job's value at an agreed milestone, common on longer builds, rather than billing the whole job at the end. Claiming a percentage that doesn't match real progress on site is a common source of cash flow trouble on staged jobs.
A change to a job's agreed scope after the quote is accepted, for example a client adding work partway through. The costly habit is doing the extra work first and raising the paperwork later, or not raising it at all, which is how scope creep eats a job's margin.
An integration that moves data, invoices, payments, contacts, between a job management tool and Xero automatically instead of someone typing it into both. It usually breaks as a partial sync: invoices go across but payment status doesn't come back, so the job system and Xero disagree about what's actually been paid.
Xero's job management product, running the back office for thousands of small NZ firms until Xero retired it in 2024. Plenty of those businesses are still on whatever they landed on as a stopgap, and haven't re-chosen since.
Using software to do a repeatable step of work without someone doing it by hand each time, from a simple rule like an email trigger up to a full custom system. The word gets used loosely: if it doesn't remove an actual manual step someone was doing, it's just software.
Software that takes a goal, decides which steps to run, and runs them, rather than following a fixed script. The useful distinction from a chatbot is that an agent can call tools, so it can look something up or change a record instead of only talking about it. In Microsoft terms, a Copilot Studio agent is the common starting point.
An open standard for connecting an AI assistant to data and tools it did not ship with. A system exposes its capabilities once as named tools, and any assistant that speaks the protocol can use them. Dataverse is an MCP server, and Copilot Studio is an MCP client.
The system side of an MCP connection: the thing that publishes tools for an assistant to call. Dataverse is one, reachable at /api/mcp on your organisation URL, and an administrator controls which clients are allowed to connect to each environment.
The model doing the reading and writing inside any AI feature. It predicts text, which is why it is good at drafting and summarising and why it will produce a confident wrong answer if the information it needs was never in front of it.
Fetching the relevant documents first, then asking the model to answer using only those. It is how an agent answers questions about your policies without the model having been trained on them. When an agent gives a wrong answer, the retrieval step is usually the problem, not the model.
Pointing an agent at a specific set of sources so its answers come from your content rather than its training data. Grounding on SharePoint inherits the permissions already on those files, which is a feature and a trap at the same time.
The unit AI usage is measured and billed in. Roughly three quarters of a word in English, counted on both what you send and what comes back. It matters because a long document pasted into every request is paid for on every request.
The instruction and context sent to a model. In a built system most of the prompt is assembled by the software rather than typed by a person, which is why "just reword the prompt" is rarely the fix when an agent misbehaves.
The standing instructions a model gets before any user message: its role, its rules, what it must refuse. This is where an agent's tone and its guardrails live, and it is the first place to look when it answers in the wrong voice.
Further training a model on your own examples so it learns a style or a narrow task. It is almost never the right first move for a small firm: grounding on your documents solves the same problem for most cases without a training run to maintain.
A fluent, confident, wrong answer. It happens when a model fills a gap instead of saying it does not know, which is why the useful test of an AI feature is what it does with a question it cannot answer.
The documented way one system lets another read or change its data. If a tool has no API, integrating it means screen scraping, file exports, or retyping, and all three cost more to run than they look.
A message one system sends the moment something happens, instead of waiting to be asked. It is the difference between a job appearing in your CRM as the form is submitted and appearing in the overnight sync.
The standard sign-in handshake that lets one app act on your behalf in another without holding your password. When an automation stops working after someone leaves, an OAuth connection tied to their account is the usual cause.
A licensed account that exists for a system rather than a person, used so an integration keeps running when staff change. It is also the cleanest way to limit what an AI agent can reach, because an agent reaches exactly what the account it signed in as reaches.
A prebuilt link between Power Platform and another system. The catch is the split between standard and premium: the connector you need is often the premium one, and that changes the licence you need.
A Power Platform environment with the extra governance controls switched on, including sharing limits, usage insights and advanced connector policies. Some administrative features, such as controlling the Dataverse MCP server through connector policy, need it.
Several makers editing the same app at once, with live presence. In canvas apps it is capped at ten makers and turns off undo, redo, search and cut while more than one person is editing.
A process note stored in Dataverse so an agent follows your rule instead of guessing at it. Where you write down the exceptions that live in someone's head, in a place software can read.
A record of every change to a system, with the ability to go back. On Power Platform it means unpacking a solution into files and committing them, which is what gives you a rollback story that a single production environment cannot.
Ask on a free 30-minute call. No prep needed, and no obligation to book anything after.