Praxismith… lessons donePraxismith… lessons doneHow AI produces an answer, why it gets things wrong with such confidence, and how to give it your own facts so it doesn’t. Then the same habits used for hiring, sales, meetings, and the daily work of a shop, a clinic or an office.
How AI produces an answer, why it makes confident mistakes, and how to ask so you get what you need.
Explain how AI writes its answers, why it makes confident mistakes, and why you give it the facts yourself.
Check an AI answer for invented names, invented links and wrong numbers before you use it at work.
Get the AI to work through a business problem one step at a time, so you can check each step before you trust the answer.
Write a request with a role, a job, the facts, the limits and the layout, so you get what you need the first time.
Turn a messy work problem into a short statement of what is actually wrong, before you ask the AI for a fix.
How to give AI your own facts: short notes it can read, versions you keep up to date, and maps of how things connect.
Tell raw numbers apart from the business reasons behind them, and give the AI both.
Keep one dated page about your business that you paste at the start of AI work and update when things change.
Give the AI plain text files, a log of its mistakes, and dated snapshots instead of designed PDFs.
Use a short background note for three everyday cases: an invoice mismatch, an angry customer on WhatsApp, and a handover.
Turn a document into a table of who does what with whom, so the AI stops joining facts that don't belong together.
Write web pages that AI search tools such as ChatGPT Search, Perplexity and Google AI Overviews can quote.
Write short knowledge cards in Google Cloud's Open Knowledge Format, each with its sources, a status, an expiry date and a named checker.
Use AI to compare job applicants against your own criteria, plan sales, decide without long meetings, keep customers, and check a business plan.
Compare CVs against a scorecard you wrote, prepare interview questions, and write kind rejection emails. You make every hiring decision.
Describe who your buyer is, list their doubts before they raise them, and write short first messages that sound like a person.
Replace long status meetings with a short written decision note, and ask the AI to argue against your plan before you commit.
Spot early signs that a customer is drifting away, and test a business plan against your own costs.
Daily office work, teaching, a clinic's paperwork, and step-by-step routines for seven kinds of business.
Summarise a long report for a decision, tidy messy spreadsheet data, and write a polite, firm email.
Set up an AI tutor that asks instead of tells, explain ideas in a student's own language, and build question papers at three levels.
Turn a doctor's discharge notes into plain words for a family, with the doctor checking every line, and know where the AI must stop.
Use one saved workflow each for marketing, sales, customer service, finance, legal, healthcare and teaching, with the right person checking the result.
Plan your next job, learn from customer reviews, learn a new subject faster, research properly, and let AI take actions only with your approval.
Match your CV to a real job post, practise a hard interview, and draft a polite counter-offer you can defend.
Sort competitor reviews on Amazon and Flipkart into clear complaints, choose one to fix, and draft listing copy around it.
Test your understanding by explaining a topic to the AI, and practise conversation in a new language.
Turn several sources into one short decision note, build a rough working version of an app idea, and set up an AI agent that asks before it acts.