Every definition is written at a sixth-grade reading level and still technically correct — with an analogy, a job-site example, and a business example for each term.
Word of the Day
“IN-fer-ens”
Inference is when an AI uses what it has already learned to answer a question, make a prediction, or complete a task.
Think of it like… Studying for a test is training. Taking the test is inference.
Why it's featured — Featured because inference is the moment AI actually does its job. Once you know this word, every conversation about speed, cost, and quality of AI makes more sense.
Terms are grouped into numbered lessons of ten. New lessons are added over time.
Lesson #1
10 terms · Nano AI-Driven Ecosystem, Agent, Multi-Agent System…
Lesson #2
10 terms · Prompt, Prompt Engineering, Training Data…
Lesson #3
9 terms · Machine Learning, Neural Network, Token…
Lesson #4
10 terms · AI Memory, Short-Term Memory, Long-Term Memory…
Lesson #5
10 terms · Bias, Explainability, AI Governance…
Lesson #6
10 terms · Agentic AI, Agentic Workflow, Goal Decomposition…
Lesson #7
10 terms · Revenue Spine, Governance Gate, Evidence Classification…
Lesson #8
10 terms · Webhook, Hook, Plugin…
Lesson #9
10 terms · Indirect Prompt Injection, Jailbreak, Least Privilege…
Lesson #10
10 terms · Stephanie.ai, Job Jacket, Daily Truth Report…
Loading your progress…
Streak & weekly recap
Loading your streak…
By category
Saved on this device. Sign in to keep it across devices.
Spaced review
Loading…
0 of 99 terms started · scheduled by difficulty and how well you recalled them.
— XP
Level 1 · Apprentice
0 of 17 badges
100 XP to level 2 · each exercise is 10 XP, each finished category is a 50 XP bonus, plus 5 XP per streak day and 25 XP per streak week.
Next up: First Nail — 1 more exercise
Loading your schedule…
Press Alt + R to jump to the search results.
99 terms found.
Pronounced AY-jent
An AI worker that has a specific job to do.
Pronounced AY-jent HAND-off
Passing a job — with its notes and context — from one agent to another, or to a person, without losing the thread.
Pronounced AY-jent loop
The repeating cycle an agent runs: look at the situation, decide one action, do it, check the result, and go again.
Pronounced AY-jent PLAN-ing
The step where an AI decides the order of its actions before it starts working, and updates that order as things change.
Pronounced ay-JEN-tik A-I
AI that can take a goal, decide the steps on its own, and carry them out instead of only answering one question at a time.
Pronounced ay-JEN-tik kun-STRUK-shun WERK-floh
Using AI agents to carry a real trade task from start to finish, not just to answer questions about it.
Pronounced ay-JEN-tik WURK-flow
A repeatable chain of steps an AI agent runs end to end, where each step's result decides what it does next.
Pronounced A-I GUV-ern-ance
The written rules for who may use AI at your organization, for what, and who is accountable when it goes wrong.
Pronounced A-I MEM-uh-ree
The information an AI tool keeps about you or your work so it doesn't start from zero every time.
Pronounced AL-go-rith-um
A list of steps that tells a computer how to solve a problem.
Pronounced A-P-I
A doorway that lets two pieces of software talk to each other.
Pronounced A-P-I END-point
One specific address in an API that does one specific job, like listing customers or creating an invoice.
Pronounced A-P-I KEE
A secret string of characters that proves which account is making a request, so a service knows who to trust and bill.
Pronounced AW-dit LOG
A tamper-resistant record of who did what, when, and to which record — including actions taken by an AI.
Pronounced aw-toh-MAY-shun
Setting up work so it happens by itself, without someone doing it each time.
Pronounced aw-TON-oh-mus AY-jent
An AI agent that can complete tasks on its own within rules you give it.
Pronounced aw-TON-uh-mee LEV-ul
How much an AI is allowed to decide and do on its own before a person has to approve the next move.
Pronounced BENCH-mark
A standard test used to compare how well different AI models perform on the same task.
Pronounced BYE-uss
When an AI's answers unfairly favor one group or view because the data it learned from leaned that way.
Pronounced kal-ih-BRAY-shun
Adjusting an AI system so its answers become more accurate and reliable.
Pronounced CHAYNJ OR-der LOOP
The full round trip of a scope change: field request, priced draft, customer approval, schedule update, and invoice — with nothing skipped.
Pronounced KLY-ent POR-tul MEER-or
The customer-facing view that shows the same job facts the internal team sees, minus anything private like margins.
Pronounced KLOHZD-loop FEED-bak EN-jin
A seven-stage cycle — detect, classify, recommend, assign, execute, verify, learn — that lets a system get better at its own job over time.
Pronounced KON-text WIN-doh
How much information an AI can hold in mind at one time.