The learning path: Beginner
Your first steps: account, environments, tables, workflows and publishing.
- 1
Your account and your organisation
You create an account in the portal, secure it with a second step, and set up an organisation. You then have three environments — Development, Acceptance and Production — each its own Flow with its own database. You need an authenticator app on your phone (for example Microsoft Authenticator or 1Password).
- 2
Your first table and workflow
In the development environment you create a table of customers and a workflow that looks them up. You see how Flow works: the explorer with your modules on the left, what you have open in the middle, and at the bottom what happens when something runs. You start in the portal, at your organisation (see 'Your account and your organisation').
- 3
Publish to acceptance
What you build in Development moves on step by step: first to Acceptance, where the business tests it, later to Production. You commit a version in Flow, click Publish and finish in the portal. You start where 'Your first table and workflow' ended.
- 4
Table from a file (CSV/Excel)
Do you already have your data in Excel, or in an export from another package? Then there is nothing to retype: Flow reads the file, proposes the columns and their kind, and puts every row in a new table. Here you use a list of customer requests. You start in the portal, at your organisation.
- 5
Columns and a status
A table is more than text: an amount is a number, a date a date, and some columns may only hold certain values. With a status column you also fix the order in which a row goes through its stages, such as new → scheduled → done. Flow guards that, and writes every status change to the audit log. You make a table of repairs for a bike shop.
- 6
A workflow that decides
A lot of work is a simple rule: below a certain amount you approve a quote yourself, above it someone has to look along. You build such a rule with an If / Else step. Each outcome finishes in an End step: successful with a result, or failed with a clear reason. You try the workflow with two examples and read what it returns.
- 7
Your first AI Colleague with a knowledge base
An AI Colleague answers questions with what you know yourselves, not with whatever is somewhere on the internet. You put that knowledge in a knowledge base: texts, manuals, terms. You make a knowledge base with the bike shop's service agreements, an agent that searches it, and you try it before anyone else gets to see it.
- 8
Chat with your AI Colleague
Every environment has a chat of its own, where colleagues talk to the AI Colleagues you publish. You put your agent in the chat (if the chat is still off, you switch it on in the same place), publish it and ask a question. With the answer you see where it searched and which document it came from. You start where 'Your first AI Colleague with a knowledge base' ended.
- 9
Runs and problems
Everyone makes a typo now and then. Flow helps you find it: Problems lists what is already wrong right now, and Runs tells you for every run what went in, what came out and where it went wrong. You build a small workflow with a mistake in it on purpose, see it fail, find out why and fix it.