Online learning paths
Learning Paths in Data, Analytics and AI
Two learning paths to help you understand and interpret in a world with AI. You learn the foundations and how it works inside, so you can work with it and see its risks too.
Start with the opening course See the learning paths
- Nine months, September to June
- One block a week, online and in Spanish
- Both start with the same course
The idea
Understand and interpret in a world with AI
More and more work will be done by an AI that someone has briefed, and what becomes scarce is knowing whether what it hands you is true. That is why every concept is learned twice: for what it is worth, and as the key to an idea in AI.
The foundations
What data is, what a model is and how you know whether it gets things right. With small numbers and business cases, no blackboard formulas.
How it works
What happens between your question and the answer, and why a flawless piece of text can still be false.
How to work with it
You brief and correct with your assistant. Reading, verifying and signing off are your job: the judgment is yours.
Its risks
Each course shows what can go wrong and the sign that gives it away, right when it comes up. And in May they get a month of their own.
Two learning paths
Choose your path
Both run for nine months, September to June, and start with the same course. What changes is the four courses from November to March and the June project.
Artificial Intelligence for Professionals
Understand what you delegate: how AI works, how to work with it and where it fails.
You work in any area, from management to finance, operations, marketing, people, legal or consulting, and you already use an AI assistant or are about to hand work to agents. Or you lead a team and have to decide what gets delegated to an AI, with how much review, and answer for the result. You want to understand AI properly without learning to code.
No coding No math background
Data Analytics with an AI Copilot
The copilot writes the code; you read it, verify it and answer for the number.
You are an analyst, a financial controller or on the business side, and you work with spreadsheets and reports. Or you already use a copilot to write queries and notebooks and want to know when to trust what it returns. Or you are going to review an agent's analytical work and need to have done some yourself to judge it.
Comfortable with spreadsheets No coding experience needed
Path 1
Artificial Intelligence for Professionals, month by month
First, understand how it works. Then work with it: brief an assistant and delegate to an agent. Then find out whether what it does is any good. With that in place, you interpret data, look at the risks and do a project of your own.
September and October · From Data to Agent
The opening course. It covers the whole route in miniature, from the data table to the agent that acts, with a single case read layer by layer.
Available Shared by both paths
November · How a Language Model Works: From Your Question to Its Answer
You follow, step by step and without formulas, what happens between your question and the answer, and see why a flawless answer can be false.
Available
December and January · Briefing Well: Instructions, Context and Documents
You move from asking to briefing: you write a complete brief, give the assistant just the context and documents it needs, and demand answers that can be checked.
Available
February · Agents: Delegating Work to an AI That Acts
What an agent can do and with which permissions, how to brief it on a long task, how to read what it has done and how much autonomy to give it.
Available
March · Does It Work? Evaluating an AI Before You Trust It
You find out whether an assistant or an agent is good enough for a task of yours before you hand it over: a pass criterion, your own set of test cases and who grades.
Available
April · Reading and Presenting Data: Charts, Metrics and Reports
You read the charts, metrics and reports you are handed with a critical eye, and present your own honestly.
Available Shared by both paths
May · AI Risks: Seeing Them, Measuring Them and Responding
You gather the risks that have come up in each course and give them a method: what can go wrong, who it affects, how likely and how serious it is, and which control fits.
Available Shared by both paths
June · Project: Your Own Work, Done with AI and Signed by You
A recurring task from your own job, redone with an assistant or an agent, tested with cases and presented with its limits.
Available
Path 2
Data Analytics with an AI Copilot, month by month
It follows the cycle of an analysis: get the data, explore it without overclaiming, model, evaluate and communicate. From November to April you rebuild, piece by piece, the agent's report you met in the opening course, with data you already have.
September and October · From Data to Agent
The opening course. It covers the whole route in miniature, from the data table to the agent that acts, with a single case read layer by layer.
Available Shared by both paths
November · Asking the Data: Tables and Queries with an AI Copilot
You read a query, brief your copilot to write it and check that it counts what it should. With a stop where most mistakes happen: the table join that duplicates rows.
Available
December and January · Statistics So You Don't Fool Yourself: Explore, Estimate and Compare
Just the statistics you need to avoid overclaiming: look at the data before you summarize it, give every figure with its margin and know whether a difference is real.
Available
February · Models That Predict: From the Straight Line to Trees
What it means for a machine to learn from data, seen in the four most widely used models. You learn to read them before you fit them.
Available
March · Evaluating a Model Honestly: Validate, Measure and Explain
You test a model without cheating: on data it has not seen, with the metric that fits the decision, and against something to compare it with.
Available
April · Reading and Presenting Data: Charts, Metrics and Reports
You read the charts, metrics and reports you are handed with a critical eye, and present your own honestly.
Available Shared by both paths
May · AI Risks: Seeing Them, Measuring Them and Responding
You gather the risks that have come up in each course and give them a method: what can go wrong, who it affects, how likely and how serious it is, and which control fits.
Available Shared by both paths
June · Project: A Full Analysis, Copiloted and Verified
An analysis of your own from start to finish: a business question answered with your copilot, verified by another route and explained to whoever decides.
Available
In common
What they share
Three courses and one way of working: brief, read, verify, correct and sign off.
September and October
"From Data to Agent", the opening course. It covers the whole route in miniature, and the other courses pick it up and go deeper.
April
"Reading and Presenting Data". It serves those who analyze and those who decide equally well.
May
"AI Risks". It gathers the risks that have come up in each course and gives them a method.
How you learn
An AI copilots the practice; the judgment is yours
You get into the substance: small numbers, examples you can follow with a pencil, and business cases.
One block a week
The opening course has eight blocks; each of the others has four. Every block comes with its lessons and its exam.
One case per course
Each course has a business case, made up for the course, that runs through it from start to finish: every block works on one part.
Copiloted practice
Every week you do a hands-on exercise with your AI assistant. What you hand in is not what the assistant did: it is your reading and your verification.
Interactive exercises
Exercises inside the lesson so you can get your hands on the idea: you join two tables and see which rows get duplicated, build a tree by hand or pick the cases to test an assistant with.
An exam with explained answers
Each block ends with an exam that explains why each answer is right or wrong.
In June, a project of your own
A task from your own job done with AI, or a full analysis with your copilot. Four weeks to frame it, do it, verify it and sign it.
Risks, built in
Every course teaches you to spot a risk
Each risk is explained by how the technology works, not as a list of fears. And in May they get a month of their own: you rank them, measure them and give each one a control.
| When | Artificial Intelligence for Professionals | Data Analytics with an AI Copilot |
|---|---|---|
| November | Believing a text because of how well it is written. | The wrong figure that looks flawless: a total inflated by a table join. |
| December and January | Pasting into a chat what should never have left the company, or accepting an answer with no source. | The figure with no margin, and the difference you find only because you kept looking. |
| February | Irreversible actions with no confirmation, and instructions hidden in what the agent reads. | The model that seems to explain but only tags along: a coefficient read as a cause. |
| March | Signing off on a system because of a brilliant demo. | The accuracy that is not true: a contaminated test, or an average that hides the group the model fails most. |
| April | The chart or report that is convincing and wrong. | The chart or report that is convincing and wrong. |
| May | All four together: trusting too much, letting the wrong things out or in, harming people with no one answerable, and breaking a rule you did not know existed. | All four together: trusting too much, letting the wrong things out or in, harming people with no one answerable, and breaking a rule you did not know existed. |
| June | Falling in love with your own result: picking only the cases that turn out well. | Searching until you find: with no plan written before you look at the data, a tireless copilot ends up proving any hypothesis right. |
FAQ
What you should know
No. They are The Singular Academy's own learning paths, not an official qualification. What you take away is what you can do by the end: for example, read an agent's trace and decide how much autonomy to give it, or get a correct figure with its margin from your copilot and verify it by another route.
In Artificial Intelligence for Professionals, no: there is no coding and you do not need a math background.
In Data Analytics with an AI Copilot you do not need to know how to code either: the copilot writes the code and you learn to read it and verify it. You do need to be comfortable with a spreadsheet.
The one you already have. No specific product is taught: for the first path any general-purpose AI assistant will do; for the second you need one that can run code on files, or a notebook environment.
The exercises never ask you to upload real people's data or confidential information to your assistant.
Yes. Each course has its own page in the Virtual Classroom and can be taken separately. Bear in mind that they all assume the opening course, "From Data to Agent": it is the best place to start.
In the four courses from November to March and in the June project. The first is for people who hand work to an AI and answer for it: how a language model works, how to brief it, how to delegate to an agent and how to evaluate it. There is no coding.
The second is for people who work with data: queries, statistics, models and their evaluation, with the code written by a copilot. The opening course and the April and May courses are the same in both.
A five-step cycle: brief, read, verify, correct and sign off. You brief and correct with the assistant; reading, verifying and signing off are yours.
From each exercise you keep your logbook: the brief, what the assistant returned and the verification record, with one figure recalculated by another route, a few rows checked by eye, an edge case tested and the source of every claim.
By invitation. The courses are available, in Spanish, in The Singular Academy's Virtual Classroom: request access from the course page or write to us.
Start with the opening course
"From Data to Agent" opens both learning paths and stands on its own. If you want to know more about either of them, write to us.