Have you been paying attention to current events recently? See how well you can do on this week’s news quiz for students.
Why the Future of Learning Starts with Building
Coding has always served two purposes: the intrinsic drive to build something, and the practical path to a lucrative career. Even the most passionate code aficionados don’t dream of variables or syntax — they want to make a website, a tool, a game. For years, the career upside was impossible to ignore. You could land a stable SWE job, bootstrap your own app, or join a buzzy startup as a first hire.
Generative AI flipped the script. AI now handles the repetitive tasks that used to define entry-level developer roles. At the same time, the barrier to entry for coding and building is lower than ever — you can spin up a working prototype with just an idea and a natural language prompt.
This shift hasn’t eliminated the desire to code, but it’s changed what and who coding is for. If you’re not learning to land a junior dev role, you’re learning to build the thing you’ve been imagining, to add a technical skill to your existing career, or to understand the tools you’re already using at work. And here’s the catch: those goals all require understanding your code, not just having code that works.
At Codecademy, all of these changes excite us about the future of learning to code. We’re introducing the AI Builder, a new project-based learning tool that flips the script by teaching you how to work with AI-generated code from the start. Our approach brings together the immediacy of modern AI tools and the rigor of real instructional design.
Why we created the AI Builder
AI’s speed and efficiency often come with a tradeoff; you can get working code immediately, but you don’t really know what it’s giving you or why it’s built a certain way. Developers use the term “vibe coding” to describe this phenomenon — it’s fast, fun, but shallow; great for demos, less great for long-term skill-building.
If your goal is to understand what you’re building, generic AI output alone won’t get you there. And the more you push these tools into real-world complexity, “the harder it is for them to give you exactly what you want,” says Zoe Bachman, Head of Learning at Codecademy.
With the AI Builder you get an education along with the AI output. In the workspace, you can toggle between two tabs: Build, where you work directly with a project and can modify and change code in real time; and Learn, where you get a personalized learning roadmap that’s based on your project.
“We pair the experience of having a working app with a learning path that allows you to reverse engineer how it’s built, so you can deeply understand it and modify it confidently,” Zoe says. We’re calling our hybrid approach to learning-driven development “vibe learning” — it’s powered by AI guidance but rooted in learning science.
Build first; learn continuously
With AI Builder, you start with what you want to do: build the thing in your head. Whether that’s a habit tracker, a portfolio site, or the seed of a bigger idea, you don’t need to have prior coding knowledge to learn and build with the AI Builder. In other words, there are no pre-requisites for creation.
You create a prototype by typing what you’d like to create in natural language. The AI chatbot will ask a few clarifying questions about your needs and overall goal before generating the project. Once the project is created, you can use the chat function to continue describing what you want. (You’re also welcome to go right into the code and start making changes if you already know your way around!)
“It was fun to build something so quickly and be able to see the code and a learning plan for it,” says Grace Krishna, a Code Crew member who beta tested the AI Builder.
When you need clarity on what’s going on behind the scenes in your code, or you hit a wall with AI, that’s a great time to flip over to the Learn tab.
We’re calling our hybrid approach to learning-driven development ‘vibe learning’ — it’s powered by AI guidance but rooted in learning science.
Your project becomes the curriculum
Rather than teaching concepts in the abstract and hoping learners translate them later, AI Builder removes that translation tax entirely. “We’re showing you specifically your code from your project and helping you understand it,” Zoe says.
To build that personalized curriculum, the AI Builder breaks your project’s code into clear milestones and tasks. For each task, it generates an interactive learning loop, which is an activity designed to help you form a mental model of what your specific code is doing.
These loops help you understand the logic behind each part of your project, so you can confidently apply the same thinking to other sections, or even future projects. This approach also ensures everything you learn is directly relevant to what you’re making — so you don’t have to guess when you’ll ever use this.
Why this is vibe learning (not vibe coding)
A key misconception about AI‑assisted development is that it makes learning superficial. AI Builder challenges that by grounding the entire experience in learning science rather than simple code generation. Our entire system is intentionally designed for you to retain knowledge. So, while it might not feel like you’re taking a course, you’re absorbing key concepts just by interacting with AI-generated code.
A Socratic AI, not an answer-spitting chatbot
Our educational AI chatbot is designed to guide you toward an answer through an in-depth questioning approach that’s based on the Socratic method of teaching. Instead of spitting out shortcuts or answers like AI typically does, you get strategic nudges, hints, and questions that build durable mental models.
Research on AI in education shows that just providing an answer makes it harder for learners to retain the information on their own. Zoe compares the Socratic AI to “a personalized tutor, facilitating you acquiring more knowledge, so you’re not totally left on your own.” Our method encourages you to think critically so you really grasp the concepts and can continue to use them in the real world.
Learning loops with real instructional design
Behind the scenes, every learning loop in the Learn tab is built on proven frameworks like inductive learning and the 5E model, a popular STEM teaching framework that’s shorthand for engage, explore, explain, elaborate, and evaluate.
You’ll notice that the questions and exercises in the Learn tab feel different than the rest of our courses and paths, and that’s intentional. “The learning loops are designed very well — they get you there inductively,” Zoe says. They’re exploratory without being overwhelming, and evaluative without feeling like tests.
Negar Vahid, a beta tester for the AI Builder appreciated the AI’s interactive question format. “The question-based learning feels engaging, and the starter project it builds is simple but useful,” she says.
This structure ensures you don’t develop the wrong mental models — a known risk in fully constructivist or student-centered environments — while still giving you the freedom to explore.
Why learn when AI can build?
There are some projects that are well-suited for simply vibe coding, like making a personal HTML website or a single-use script to automate a one-time task. Tools like Lovable and v0 are suited exactly for these types of projects.
The longer your code needs to live, and the more complex your project becomes, the more you need to actually understand what you’re building. Joe Holmes, Codecademy Curriculum Developer in the AI and machine learning domain, uses the term “ignorance debt” to describe what happens when you don’t:
“It’s like tech debt squared. It’s much, much worse,” Joe says. “You don’t know what kind of code is coming out. You just are only looking at: Does this kind of generally appear to be what I asked for? You don’t know if there are security flaws. You don’t know if there are performance flaws. You don’t know if you’re leaking sensitive information. You don’t know how to fix anything.”
The tipping point comes down to two factors: complexity and time. If you’re developing software professionally, you’re legally responsible for the code you output. If you’re building something that will serve actual users, you need to be accountable for security, performance, and maintainability. And if your project will need updates or fixes over time (which most do) understanding your codebase becomes essential, not optional.
The good news? Learning doesn’t have to feel like eating your vegetables. “Kids hate veggies and broccoli because we don’t cook it well enough to make it tasty when we first introduce it to them,” says Nhi Pham, Codecademy Curriculum Developer. The same is true for teaching AI: “If you do it well, you’re inspiring people to have these very hygienic practices when working with AI,” she says.
That’s exactly what AI Builder is designed to do — make learning feel as immediate and rewarding as building, so you develop good habits from the start rather than building a lifelong aversion to understanding your own code.
Get started with the AI Builder
AI isn’t a replacement for learning, it’s a tool — and a powerful one when it comes to education. Our new AI Builder allows for “just‑in‑time learning that’s highly personalized,” Zoe says. Even the best teachers or bootcamps can’t deliver that for every learner, on every project, instantly. Perhaps the most exciting vision is how AI changes what a learning environment can be.
Zoe described it beautifully: “I imagine the AI Builder as a workspace… like having all your resources around you and an AI tutor in the background.”
That’s the shift: from learning before you build to learning while you build. We can’t wait to see what you create.
Why the Future of Learning Starts with Building
Coding has always served two purposes: the intrinsic drive to build something, and the practical path to a lucrative career. Even the most passionate code aficionados don’t dream of variables or syntax — they want to make a website, a tool, a game. For years, the career upside was impossible to ignore. You could land a stable SWE job, bootstrap your own app, or join a buzzy startup as a first hire.
Generative AI flipped the script. AI now handles the repetitive tasks that used to define entry-level developer roles. At the same time, the barrier to entry for coding and building is lower than ever — you can spin up a working prototype with just an idea and a natural language prompt.
This shift hasn’t eliminated the desire to code, but it’s changed what and who coding is for. If you’re not learning to land a junior dev role, you’re learning to build the thing you’ve been imagining, to add a technical skill to your existing career, or to understand the tools you’re already using at work. And here’s the catch: those goals all require understanding your code, not just having code that works.
At Codecademy, all of these changes excite us about the future of learning to code. We’re introducing the AI Builder, a new project-based learning tool that flips the script by teaching you how to work with AI-generated code from the start. Our approach brings together the immediacy of modern AI tools and the rigor of real instructional design.
Why we created the AI Builder
AI’s speed and efficiency often come with a tradeoff; you can get working code immediately, but you don’t really know what it’s giving you or why it’s built a certain way. Developers use the term “vibe coding” to describe this phenomenon — it’s fast, fun, but shallow; great for demos, less great for long-term skill-building.
If your goal is to understand what you’re building, generic AI output alone won’t get you there. And the more you push these tools into real-world complexity, “the harder it is for them to give you exactly what you want,” says Zoe Bachman, Head of Learning at Codecademy.
With the AI Builder you get an education along with the AI output. In the workspace, you can toggle between two tabs: Build, where you work directly with a project and can modify and change code in real time; and Learn, where you get a personalized learning roadmap that’s based on your project.
“We pair the experience of having a working app with a learning path that allows you to reverse engineer how it’s built, so you can deeply understand it and modify it confidently,” Zoe says. We’re calling our hybrid approach to learning-driven development “vibe learning” — it’s powered by AI guidance but rooted in learning science.
Build first; learn continuously
With AI Builder, you start with what you want to do: build the thing in your head. Whether that’s a habit tracker, a portfolio site, or the seed of a bigger idea, you don’t need to have prior coding knowledge to learn and build with the AI Builder. In other words, there are no pre-requisites for creation.
You create a prototype by typing what you’d like to create in natural language. The AI chatbot will ask a few clarifying questions about your needs and overall goal before generating the project. Once the project is created, you can use the chat function to continue describing what you want. (You’re also welcome to go right into the code and start making changes if you already know your way around!)
“It was fun to build something so quickly and be able to see the code and a learning plan for it,” says Grace Krishna, a Code Crew member who beta tested the AI Builder.
When you need clarity on what’s going on behind the scenes in your code, or you hit a wall with AI, that’s a great time to flip over to the Learn tab.
We’re calling our hybrid approach to learning-driven development ‘vibe learning’ — it’s powered by AI guidance but rooted in learning science.
Your project becomes the curriculum
Rather than teaching concepts in the abstract and hoping learners translate them later, AI Builder removes that translation tax entirely. “We’re showing you specifically your code from your project and helping you understand it,” Zoe says.
To build that personalized curriculum, the AI Builder breaks your project’s code into clear milestones and tasks. For each task, it generates an interactive learning loop, which is an activity designed to help you form a mental model of what your specific code is doing.
These loops help you understand the logic behind each part of your project, so you can confidently apply the same thinking to other sections, or even future projects. This approach also ensures everything you learn is directly relevant to what you’re making — so you don’t have to guess when you’ll ever use this.
Why this is vibe learning (not vibe coding)
A key misconception about AI‑assisted development is that it makes learning superficial. AI Builder challenges that by grounding the entire experience in learning science rather than simple code generation. Our entire system is intentionally designed for you to retain knowledge. So, while it might not feel like you’re taking a course, you’re absorbing key concepts just by interacting with AI-generated code.
A Socratic AI, not an answer-spitting chatbot
Our educational AI chatbot is designed to guide you toward an answer through an in-depth questioning approach that’s based on the Socratic method of teaching. Instead of spitting out shortcuts or answers like AI typically does, you get strategic nudges, hints, and questions that build durable mental models.
Research on AI in education shows that just providing an answer makes it harder for learners to retain the information on their own. Zoe compares the Socratic AI to “a personalized tutor, facilitating you acquiring more knowledge, so you’re not totally left on your own.” Our method encourages you to think critically so you really grasp the concepts and can continue to use them in the real world.
Learning loops with real instructional design
Behind the scenes, every learning loop in the Learn tab is built on proven frameworks like inductive learning and the 5E model, a popular STEM teaching framework that’s shorthand for engage, explore, explain, elaborate, and evaluate.
You’ll notice that the questions and exercises in the Learn tab feel different than the rest of our courses and paths, and that’s intentional. “The learning loops are designed very well — they get you there inductively,” Zoe says. They’re exploratory without being overwhelming, and evaluative without feeling like tests.
Negar Vahid, a beta tester for the AI Builder appreciated the AI’s interactive question format. “The question-based learning feels engaging, and the starter project it builds is simple but useful,” she says.
This structure ensures you don’t develop the wrong mental models — a known risk in fully constructivist or student-centered environments — while still giving you the freedom to explore.
Why learn when AI can build?
There are some projects that are well-suited for simply vibe coding, like making a personal HTML website or a single-use script to automate a one-time task. Tools like Lovable and v0 are suited exactly for these types of projects.
The longer your code needs to live, and the more complex your project becomes, the more you need to actually understand what you’re building. Joe Holmes, Codecademy Curriculum Developer in the AI and machine learning domain, uses the term “ignorance debt” to describe what happens when you don’t:
“It’s like tech debt squared. It’s much, much worse,” Joe says. “You don’t know what kind of code is coming out. You just are only looking at: Does this kind of generally appear to be what I asked for? You don’t know if there are security flaws. You don’t know if there are performance flaws. You don’t know if you’re leaking sensitive information. You don’t know how to fix anything.”
The tipping point comes down to two factors: complexity and time. If you’re developing software professionally, you’re legally responsible for the code you output. If you’re building something that will serve actual users, you need to be accountable for security, performance, and maintainability. And if your project will need updates or fixes over time (which most do) understanding your codebase becomes essential, not optional.
The good news? Learning doesn’t have to feel like eating your vegetables. “Kids hate veggies and broccoli because we don’t cook it well enough to make it tasty when we first introduce it to them,” says Nhi Pham, Codecademy Curriculum Developer. The same is true for teaching AI: “If you do it well, you’re inspiring people to have these very hygienic practices when working with AI,” she says.
That’s exactly what AI Builder is designed to do — make learning feel as immediate and rewarding as building, so you develop good habits from the start rather than building a lifelong aversion to understanding your own code.
Get started with the AI Builder
AI isn’t a replacement for learning, it’s a tool — and a powerful one when it comes to education. Our new AI Builder allows for “just‑in‑time learning that’s highly personalized,” Zoe says. Even the best teachers or bootcamps can’t deliver that for every learner, on every project, instantly. Perhaps the most exciting vision is how AI changes what a learning environment can be.
Zoe described it beautifully: “I imagine the AI Builder as a workspace… like having all your resources around you and an AI tutor in the background.”
That’s the shift: from learning before you build to learning while you build. We can’t wait to see what you create.
Why the Future of Learning Starts with Building
Coding has always served two purposes: the intrinsic drive to build something, and the practical path to a lucrative career. Even the most passionate code aficionados don’t dream of variables or syntax — they want to make a website, a tool, a game. For years, the career upside was impossible to ignore. You could land a stable SWE job, bootstrap your own app, or join a buzzy startup as a first hire.
Generative AI flipped the script. AI now handles the repetitive tasks that used to define entry-level developer roles. At the same time, the barrier to entry for coding and building is lower than ever — you can spin up a working prototype with just an idea and a natural language prompt.
This shift hasn’t eliminated the desire to code, but it’s changed what and who coding is for. If you’re not learning to land a junior dev role, you’re learning to build the thing you’ve been imagining, to add a technical skill to your existing career, or to understand the tools you’re already using at work. And here’s the catch: those goals all require understanding your code, not just having code that works.
At Codecademy, all of these changes excite us about the future of learning to code. We’re introducing the AI Builder, a new project-based learning tool that flips the script by teaching you how to work with AI-generated code from the start. Our approach brings together the immediacy of modern AI tools and the rigor of real instructional design.
Why we created the AI Builder
AI’s speed and efficiency often come with a tradeoff; you can get working code immediately, but you don’t really know what it’s giving you or why it’s built a certain way. Developers use the term “vibe coding” to describe this phenomenon — it’s fast, fun, but shallow; great for demos, less great for long-term skill-building.
If your goal is to understand what you’re building, generic AI output alone won’t get you there. And the more you push these tools into real-world complexity, “the harder it is for them to give you exactly what you want,” says Zoe Bachman, Head of Learning at Codecademy.
With the AI Builder you get an education along with the AI output. In the workspace, you can toggle between two tabs: Build, where you work directly with a project and can modify and change code in real time; and Learn, where you get a personalized learning roadmap that’s based on your project.
“We pair the experience of having a working app with a learning path that allows you to reverse engineer how it’s built, so you can deeply understand it and modify it confidently,” Zoe says. We’re calling our hybrid approach to learning-driven development “vibe learning” — it’s powered by AI guidance but rooted in learning science.
Build first; learn continuously
With AI Builder, you start with what you want to do: build the thing in your head. Whether that’s a habit tracker, a portfolio site, or the seed of a bigger idea, you don’t need to have prior coding knowledge to learn and build with the AI Builder. In other words, there are no pre-requisites for creation.
You create a prototype by typing what you’d like to create in natural language. The AI chatbot will ask a few clarifying questions about your needs and overall goal before generating the project. Once the project is created, you can use the chat function to continue describing what you want. (You’re also welcome to go right into the code and start making changes if you already know your way around!)
“It was fun to build something so quickly and be able to see the code and a learning plan for it,” says Grace Krishna, a Code Crew member who beta tested the AI Builder.
When you need clarity on what’s going on behind the scenes in your code, or you hit a wall with AI, that’s a great time to flip over to the Learn tab.
We’re calling our hybrid approach to learning-driven development ‘vibe learning’ — it’s powered by AI guidance but rooted in learning science.
Your project becomes the curriculum
Rather than teaching concepts in the abstract and hoping learners translate them later, AI Builder removes that translation tax entirely. “We’re showing you specifically your code from your project and helping you understand it,” Zoe says.
To build that personalized curriculum, the AI Builder breaks your project’s code into clear milestones and tasks. For each task, it generates an interactive learning loop, which is an activity designed to help you form a mental model of what your specific code is doing.
These loops help you understand the logic behind each part of your project, so you can confidently apply the same thinking to other sections, or even future projects. This approach also ensures everything you learn is directly relevant to what you’re making — so you don’t have to guess when you’ll ever use this.
Why this is vibe learning (not vibe coding)
A key misconception about AI‑assisted development is that it makes learning superficial. AI Builder challenges that by grounding the entire experience in learning science rather than simple code generation. Our entire system is intentionally designed for you to retain knowledge. So, while it might not feel like you’re taking a course, you’re absorbing key concepts just by interacting with AI-generated code.
A Socratic AI, not an answer-spitting chatbot
Our educational AI chatbot is designed to guide you toward an answer through an in-depth questioning approach that’s based on the Socratic method of teaching. Instead of spitting out shortcuts or answers like AI typically does, you get strategic nudges, hints, and questions that build durable mental models.
Research on AI in education shows that just providing an answer makes it harder for learners to retain the information on their own. Zoe compares the Socratic AI to “a personalized tutor, facilitating you acquiring more knowledge, so you’re not totally left on your own.” Our method encourages you to think critically so you really grasp the concepts and can continue to use them in the real world.
Learning loops with real instructional design
Behind the scenes, every learning loop in the Learn tab is built on proven frameworks like inductive learning and the 5E model, a popular STEM teaching framework that’s shorthand for engage, explore, explain, elaborate, and evaluate.
You’ll notice that the questions and exercises in the Learn tab feel different than the rest of our courses and paths, and that’s intentional. “The learning loops are designed very well — they get you there inductively,” Zoe says. They’re exploratory without being overwhelming, and evaluative without feeling like tests.
Negar Vahid, a beta tester for the AI Builder appreciated the AI’s interactive question format. “The question-based learning feels engaging, and the starter project it builds is simple but useful,” she says.
This structure ensures you don’t develop the wrong mental models — a known risk in fully constructivist or student-centered environments — while still giving you the freedom to explore.
Why learn when AI can build?
There are some projects that are well-suited for simply vibe coding, like making a personal HTML website or a single-use script to automate a one-time task. Tools like Lovable and v0 are suited exactly for these types of projects.
The longer your code needs to live, and the more complex your project becomes, the more you need to actually understand what you’re building. Joe Holmes, Codecademy Curriculum Developer in the AI and machine learning domain, uses the term “ignorance debt” to describe what happens when you don’t:
“It’s like tech debt squared. It’s much, much worse,” Joe says. “You don’t know what kind of code is coming out. You just are only looking at: Does this kind of generally appear to be what I asked for? You don’t know if there are security flaws. You don’t know if there are performance flaws. You don’t know if you’re leaking sensitive information. You don’t know how to fix anything.”
The tipping point comes down to two factors: complexity and time. If you’re developing software professionally, you’re legally responsible for the code you output. If you’re building something that will serve actual users, you need to be accountable for security, performance, and maintainability. And if your project will need updates or fixes over time (which most do) understanding your codebase becomes essential, not optional.
The good news? Learning doesn’t have to feel like eating your vegetables. “Kids hate veggies and broccoli because we don’t cook it well enough to make it tasty when we first introduce it to them,” says Nhi Pham, Codecademy Curriculum Developer. The same is true for teaching AI: “If you do it well, you’re inspiring people to have these very hygienic practices when working with AI,” she says.
That’s exactly what AI Builder is designed to do — make learning feel as immediate and rewarding as building, so you develop good habits from the start rather than building a lifelong aversion to understanding your own code.
Get started with the AI Builder
AI isn’t a replacement for learning, it’s a tool — and a powerful one when it comes to education. Our new AI Builder allows for “just‑in‑time learning that’s highly personalized,” Zoe says. Even the best teachers or bootcamps can’t deliver that for every learner, on every project, instantly. Perhaps the most exciting vision is how AI changes what a learning environment can be.
Zoe described it beautifully: “I imagine the AI Builder as a workspace… like having all your resources around you and an AI tutor in the background.”
That’s the shift: from learning before you build to learning while you build. We can’t wait to see what you create.
Why the Future of Learning Starts with Building
Coding has always served two purposes: the intrinsic drive to build something, and the practical path to a lucrative career. Even the most passionate code aficionados don’t dream of variables or syntax — they want to make a website, a tool, a game. For years, the career upside was impossible to ignore. You could land a stable SWE job, bootstrap your own app, or join a buzzy startup as a first hire.
Generative AI flipped the script. AI now handles the repetitive tasks that used to define entry-level developer roles. At the same time, the barrier to entry for coding and building is lower than ever — you can spin up a working prototype with just an idea and a natural language prompt.
This shift hasn’t eliminated the desire to code, but it’s changed what and who coding is for. If you’re not learning to land a junior dev role, you’re learning to build the thing you’ve been imagining, to add a technical skill to your existing career, or to understand the tools you’re already using at work. And here’s the catch: those goals all require understanding your code, not just having code that works.
At Codecademy, all of these changes excite us about the future of learning to code. We’re introducing the AI Builder, a new project-based learning tool that flips the script by teaching you how to work with AI-generated code from the start. Our approach brings together the immediacy of modern AI tools and the rigor of real instructional design.
Why we created the AI Builder
AI’s speed and efficiency often come with a tradeoff; you can get working code immediately, but you don’t really know what it’s giving you or why it’s built a certain way. Developers use the term “vibe coding” to describe this phenomenon — it’s fast, fun, but shallow; great for demos, less great for long-term skill-building.
If your goal is to understand what you’re building, generic AI output alone won’t get you there. And the more you push these tools into real-world complexity, “the harder it is for them to give you exactly what you want,” says Zoe Bachman, Head of Learning at Codecademy.
With the AI Builder you get an education along with the AI output. In the workspace, you can toggle between two tabs: Build, where you work directly with a project and can modify and change code in real time; and Learn, where you get a personalized learning roadmap that’s based on your project.
“We pair the experience of having a working app with a learning path that allows you to reverse engineer how it’s built, so you can deeply understand it and modify it confidently,” Zoe says. We’re calling our hybrid approach to learning-driven development “vibe learning” — it’s powered by AI guidance but rooted in learning science.
Build first; learn continuously
With AI Builder, you start with what you want to do: build the thing in your head. Whether that’s a habit tracker, a portfolio site, or the seed of a bigger idea, you don’t need to have prior coding knowledge to learn and build with the AI Builder. In other words, there are no pre-requisites for creation.
You create a prototype by typing what you’d like to create in natural language. The AI chatbot will ask a few clarifying questions about your needs and overall goal before generating the project. Once the project is created, you can use the chat function to continue describing what you want. (You’re also welcome to go right into the code and start making changes if you already know your way around!)
“It was fun to build something so quickly and be able to see the code and a learning plan for it,” says Grace Krishna, a Code Crew member who beta tested the AI Builder.
When you need clarity on what’s going on behind the scenes in your code, or you hit a wall with AI, that’s a great time to flip over to the Learn tab.
We’re calling our hybrid approach to learning-driven development ‘vibe learning’ — it’s powered by AI guidance but rooted in learning science.
Your project becomes the curriculum
Rather than teaching concepts in the abstract and hoping learners translate them later, AI Builder removes that translation tax entirely. “We’re showing you specifically your code from your project and helping you understand it,” Zoe says.
To build that personalized curriculum, the AI Builder breaks your project’s code into clear milestones and tasks. For each task, it generates an interactive learning loop, which is an activity designed to help you form a mental model of what your specific code is doing.
These loops help you understand the logic behind each part of your project, so you can confidently apply the same thinking to other sections, or even future projects. This approach also ensures everything you learn is directly relevant to what you’re making — so you don’t have to guess when you’ll ever use this.
Why this is vibe learning (not vibe coding)
A key misconception about AI‑assisted development is that it makes learning superficial. AI Builder challenges that by grounding the entire experience in learning science rather than simple code generation. Our entire system is intentionally designed for you to retain knowledge. So, while it might not feel like you’re taking a course, you’re absorbing key concepts just by interacting with AI-generated code.
A Socratic AI, not an answer-spitting chatbot
Our educational AI chatbot is designed to guide you toward an answer through an in-depth questioning approach that’s based on the Socratic method of teaching. Instead of spitting out shortcuts or answers like AI typically does, you get strategic nudges, hints, and questions that build durable mental models.
Research on AI in education shows that just providing an answer makes it harder for learners to retain the information on their own. Zoe compares the Socratic AI to “a personalized tutor, facilitating you acquiring more knowledge, so you’re not totally left on your own.” Our method encourages you to think critically so you really grasp the concepts and can continue to use them in the real world.
Learning loops with real instructional design
Behind the scenes, every learning loop in the Learn tab is built on proven frameworks like inductive learning and the 5E model, a popular STEM teaching framework that’s shorthand for engage, explore, explain, elaborate, and evaluate.
You’ll notice that the questions and exercises in the Learn tab feel different than the rest of our courses and paths, and that’s intentional. “The learning loops are designed very well — they get you there inductively,” Zoe says. They’re exploratory without being overwhelming, and evaluative without feeling like tests.
Negar Vahid, a beta tester for the AI Builder appreciated the AI’s interactive question format. “The question-based learning feels engaging, and the starter project it builds is simple but useful,” she says.
This structure ensures you don’t develop the wrong mental models — a known risk in fully constructivist or student-centered environments — while still giving you the freedom to explore.
Why learn when AI can build?
There are some projects that are well-suited for simply vibe coding, like making a personal HTML website or a single-use script to automate a one-time task. Tools like Lovable and v0 are suited exactly for these types of projects.
The longer your code needs to live, and the more complex your project becomes, the more you need to actually understand what you’re building. Joe Holmes, Codecademy Curriculum Developer in the AI and machine learning domain, uses the term “ignorance debt” to describe what happens when you don’t:
“It’s like tech debt squared. It’s much, much worse,” Joe says. “You don’t know what kind of code is coming out. You just are only looking at: Does this kind of generally appear to be what I asked for? You don’t know if there are security flaws. You don’t know if there are performance flaws. You don’t know if you’re leaking sensitive information. You don’t know how to fix anything.”
The tipping point comes down to two factors: complexity and time. If you’re developing software professionally, you’re legally responsible for the code you output. If you’re building something that will serve actual users, you need to be accountable for security, performance, and maintainability. And if your project will need updates or fixes over time (which most do) understanding your codebase becomes essential, not optional.
The good news? Learning doesn’t have to feel like eating your vegetables. “Kids hate veggies and broccoli because we don’t cook it well enough to make it tasty when we first introduce it to them,” says Nhi Pham, Codecademy Curriculum Developer. The same is true for teaching AI: “If you do it well, you’re inspiring people to have these very hygienic practices when working with AI,” she says.
That’s exactly what AI Builder is designed to do — make learning feel as immediate and rewarding as building, so you develop good habits from the start rather than building a lifelong aversion to understanding your own code.
Get started with the AI Builder
AI isn’t a replacement for learning, it’s a tool — and a powerful one when it comes to education. Our new AI Builder allows for “just‑in‑time learning that’s highly personalized,” Zoe says. Even the best teachers or bootcamps can’t deliver that for every learner, on every project, instantly. Perhaps the most exciting vision is how AI changes what a learning environment can be.
Zoe described it beautifully: “I imagine the AI Builder as a workspace… like having all your resources around you and an AI tutor in the background.”
That’s the shift: from learning before you build to learning while you build. We can’t wait to see what you create.
AI Inudation – Know the Questions to ask
It’s that time of year when folks flock to trade shows like LTUK and ATD.
A quick perusal of exhibitors shows many with some form of AI.
AI video creators. AI transformation of content.
AI is everywhere in these shows—powering scenarios, skill development, and more.
To help you cut through the noise and ask the right questions about AI at these events, here’s a focused guide (this post).
Terms to Ask
MCP (Model Context Protocol) is the latest; go ahead and ask whether they are using it.
Other MCP Stuff to Ask
- Can I have the platform/knowledge management (i.e., the vendor you are talking to)- be used in your LLM – your company’s one, if they are using one.
It works this way – you are currently using your own Learning Language Model – let’s say, Copilot or Claude.
The vendor’s platform using MCP can be integrated into the LLM, with some capabilities. I have yet to see a 100% system MCP integrated into your own LLM.
For example, a vendor-knowledge management system, Talvi (product review next week), can do this – MCP.
2. Does your system have an MCP store? This is the latest, and if you need it or seek something like it, it is a huge plus for a vendor to have it. There should be more than 10 MCP offerings
Vendors love to talk about guardrails, and they have a RAG.
The premise is simple: we have all this stuff, so the human loop element isn’t as crucial, since we are 99.99% accurate or fully accurate with your own content.
However, these claims are not entirely accurate.
If they are using an AWS solution, such as Bedrock, it would make sense to ask how they use it in your system.
If a vendor says they are using Claude, then I want to know how they handle the data throttle that Anthropic is doing with mass usage – yep, that’s right, just like those telecomms.
If your company values environmental protection, ask about data centers’ impact—energy use, costs, and water consumption.
I want to know whether the vendor is using or planning to use an SLM (the term varies, but it means a small language model that can run on a mobile device, even a laptop, and push out nearly as many parameters as the big LLMs).
Here are a few other items I would want to know
- Token fees are costs for using an AI system, as tokens are units of text processed. Low usage poses little concern, but high usage raises fees. For example, 10,000 users a day asking the AI Assistant questions increases fees, since each prompt is not free.
- Natural Language – I want to learn more about their approach and ongoing work in Natural Language Processing (which you want).
- Processing chips are specialized hardware needed for AI operations. A major shortage of these chips drives up vendor costs, since LLMs or SLMs—foundations for Generative AI (Gen AI)—must be used. NVIDIA, the leading supplier, faces this challenge. Even if a vendor builds its own LLM, significant computational power is still necessary. Vendors pass these higher costs to you. What strategy does the vendor have?
- Does the AI assistant’s prompt window offer suggestions? You want this, as suggestions spur new ideas or actions.
- Does the vendor display a warning: ‘AI can make mistakes. Always verify accuracy before accepting results.’ Some suggest human review before accepting.
- Does the system allow you to create your own AI agent (a vendor may call it Agentic)? The creation capability has to be easy to use. I’ve seen ones that are very complex.
Vendors often assume people know AI can make mistakes and compare it to human error.
I rely on AI, expecting 100% accuracy, but even small mistakes, like in PowerPoint, can hurt a company. AI bias can slip in unchecked.
There is data showing the potential impact on folks who use AI a lot and get attached to it, as if it can do everything for them.
I laugh at claims that ‘AI frees employees for other work.’
If AI handles tasks, why would employees take on others if they see AI doing them?
While AI cannot do deep thinking, it can handle many tasks in L&D, Training, Marketing, and Finance.
You need a spreadsheet with computations based on a series of data points, usually handled by macros?
Gen AI can handle those spreadsheet needs.
What I want to know
Ask vendors how their Gen AI platform enables you to measure and enhance the Impact of Learning, which is a more meaningful goal than traditional ROI.
I’m a huge believer in IOL over ROI, because I can tie the impact to the company’s business goals and objectives.
Use data instead of vague claims like ‘happy employees work more.’ Ignore unknown variables.
Show the impact on goals or profit centers; usually, better results lead to more budget. Online learning isn’t one-and-done.
Show your boss external training generated $250K—big impact, clear benefits.
Benefits include more budget, resources, and less oversight from HR/IT (if applicable).
Now, let’s consider your specific objectives:
- The vendor must provide ongoing AI literacy for all users. If sales lack AI knowledge, why buy? Vendors should update, not you. Stay informed and keep your team updated, as few read support guides. Some vendors shift responsibility to buyers.
- Ask questions. Don’t accept jargon; clarify if unsure.
- AI roleplaying can bore users. Roleplays should require deep thought and align with job roles. If in construction, I’d prefer a relatable roleplay scenario.
The fact of the matter is that people learn better when the stakeholder is like them.
Thus, the coach isn’t a manager – rather, it is a person who works in, say, construction and achieved XAB, which is why they are a coach for that employee.
Transparency
I am a huge fan of Bedrock and the other AWS AI offerings.
So much so that, for my FindAnLMS platform, I will be using an AWS AI product (launch in Q4, 2026).
Bottom Line
DigitalChalk is on track to become the first LMS with a complete learning intelligence infrastructure built 100% on AI.
What I have seen is really getting me thrilled about what is possible and what is on the map with DigitalChalk.
There are other vendors that use Bedrock, such as SparkLearn, designed for frontline workers, and Cornerstone, which works with AWS on a variety of AI offerings.
If you want to see who I have in my top tAI learning systems (as of May 2026) (rankings and insight to be published), Thrive Learning is definitely impressing me. DigitalChalk, Absorb, Cornerstone, and Talvi are equally worth a check.
Learn Amp is another one, too.
They are ideal for L&D, IMO, and their AI-based learning is strong.
At the end of all those AI learning tech offerings, ask yourself two questions:
Ask: How does this solution help validate the company’s meaningful learning impact?
Ask: Will using this AI system drive repeat engagement and support ongoing learning for your users?
Remember: The true goal is to entice learners—your end users—to return to the system because the AI makes it valuable to them.
Not you.
E-Learning 24/7
Why the Future of Learning Starts with Building
Coding has always served two purposes: the intrinsic drive to build something, and the practical path to a lucrative career. Even the most passionate code aficionados don’t dream of variables or syntax — they want to make a website, a tool, a game. For years, the career upside was impossible to ignore. You could land a stable SWE job, bootstrap your own app, or join a buzzy startup as a first hire.
Generative AI flipped the script. AI now handles the repetitive tasks that used to define entry-level developer roles. At the same time, the barrier to entry for coding and building is lower than ever — you can spin up a working prototype with just an idea and a natural language prompt.
This shift hasn’t eliminated the desire to code, but it’s changed what and who coding is for. If you’re not learning to land a junior dev role, you’re learning to build the thing you’ve been imagining, to add a technical skill to your existing career, or to understand the tools you’re already using at work. And here’s the catch: those goals all require understanding your code, not just having code that works.
At Codecademy, all of these changes excite us about the future of learning to code. We’re introducing the AI Builder, a new project-based learning tool that flips the script by teaching you how to work with AI-generated code from the start. Our approach brings together the immediacy of modern AI tools and the rigor of real instructional design.
Why we created the AI Builder
AI’s speed and efficiency often come with a tradeoff; you can get working code immediately, but you don’t really know what it’s giving you or why it’s built a certain way. Developers use the term “vibe coding” to describe this phenomenon — it’s fast, fun, but shallow; great for demos, less great for long-term skill-building.
If your goal is to understand what you’re building, generic AI output alone won’t get you there. And the more you push these tools into real-world complexity, “the harder it is for them to give you exactly what you want,” says Zoe Bachman, Head of Learning at Codecademy.
With the AI Builder you get an education along with the AI output. In the workspace, you can toggle between two tabs: Build, where you work directly with a project and can modify and change code in real time; and Learn, where you get a personalized learning roadmap that’s based on your project.
“We pair the experience of having a working app with a learning path that allows you to reverse engineer how it’s built, so you can deeply understand it and modify it confidently,” Zoe says. We’re calling our hybrid approach to learning-driven development “vibe learning” — it’s powered by AI guidance but rooted in learning science.
Build first; learn continuously
With AI Builder, you start with what you want to do: build the thing in your head. Whether that’s a habit tracker, a portfolio site, or the seed of a bigger idea, you don’t need to have prior coding knowledge to learn and build with the AI Builder. In other words, there are no pre-requisites for creation.
You create a prototype by typing what you’d like to create in natural language. The AI chatbot will ask a few clarifying questions about your needs and overall goal before generating the project. Once the project is created, you can use the chat function to continue describing what you want. (You’re also welcome to go right into the code and start making changes if you already know your way around!)
“It was fun to build something so quickly and be able to see the code and a learning plan for it,” says Grace Krishna, a Code Crew member who beta tested the AI Builder.
When you need clarity on what’s going on behind the scenes in your code, or you hit a wall with AI, that’s a great time to flip over to the Learn tab.
We’re calling our hybrid approach to learning-driven development ‘vibe learning’ — it’s powered by AI guidance but rooted in learning science.
Your project becomes the curriculum
Rather than teaching concepts in the abstract and hoping learners translate them later, AI Builder removes that translation tax entirely. “We’re showing you specifically your code from your project and helping you understand it,” Zoe says.
To build that personalized curriculum, the AI Builder breaks your project’s code into clear milestones and tasks. For each task, it generates an interactive learning loop, which is an activity designed to help you form a mental model of what your specific code is doing.
These loops help you understand the logic behind each part of your project, so you can confidently apply the same thinking to other sections, or even future projects. This approach also ensures everything you learn is directly relevant to what you’re making — so you don’t have to guess when you’ll ever use this.
Why this is vibe learning (not vibe coding)
A key misconception about AI‑assisted development is that it makes learning superficial. AI Builder challenges that by grounding the entire experience in learning science rather than simple code generation. Our entire system is intentionally designed for you to retain knowledge. So, while it might not feel like you’re taking a course, you’re absorbing key concepts just by interacting with AI-generated code.
A Socratic AI, not an answer-spitting chatbot
Our educational AI chatbot is designed to guide you toward an answer through an in-depth questioning approach that’s based on the Socratic method of teaching. Instead of spitting out shortcuts or answers like AI typically does, you get strategic nudges, hints, and questions that build durable mental models.
Research on AI in education shows that just providing an answer makes it harder for learners to retain the information on their own. Zoe compares the Socratic AI to “a personalized tutor, facilitating you acquiring more knowledge, so you’re not totally left on your own.” Our method encourages you to think critically so you really grasp the concepts and can continue to use them in the real world.
Learning loops with real instructional design
Behind the scenes, every learning loop in the Learn tab is built on proven frameworks like inductive learning and the 5E model, a popular STEM teaching framework that’s shorthand for engage, explore, explain, elaborate, and evaluate.
You’ll notice that the questions and exercises in the Learn tab feel different than the rest of our courses and paths, and that’s intentional. “The learning loops are designed very well — they get you there inductively,” Zoe says. They’re exploratory without being overwhelming, and evaluative without feeling like tests.
Negar Vahid, a beta tester for the AI Builder appreciated the AI’s interactive question format. “The question-based learning feels engaging, and the starter project it builds is simple but useful,” she says.
This structure ensures you don’t develop the wrong mental models — a known risk in fully constructivist or student-centered environments — while still giving you the freedom to explore.
Why learn when AI can build?
There are some projects that are well-suited for simply vibe coding, like making a personal HTML website or a single-use script to automate a one-time task. Tools like Lovable and v0 are suited exactly for these types of projects.
The longer your code needs to live, and the more complex your project becomes, the more you need to actually understand what you’re building. Joe Holmes, Codecademy Curriculum Developer in the AI and machine learning domain, uses the term “ignorance debt” to describe what happens when you don’t:
“It’s like tech debt squared. It’s much, much worse,” Joe says. “You don’t know what kind of code is coming out. You just are only looking at: Does this kind of generally appear to be what I asked for? You don’t know if there are security flaws. You don’t know if there are performance flaws. You don’t know if you’re leaking sensitive information. You don’t know how to fix anything.”
The tipping point comes down to two factors: complexity and time. If you’re developing software professionally, you’re legally responsible for the code you output. If you’re building something that will serve actual users, you need to be accountable for security, performance, and maintainability. And if your project will need updates or fixes over time (which most do) understanding your codebase becomes essential, not optional.
The good news? Learning doesn’t have to feel like eating your vegetables. “Kids hate veggies and broccoli because we don’t cook it well enough to make it tasty when we first introduce it to them,” says Nhi Pham, Codecademy Curriculum Developer. The same is true for teaching AI: “If you do it well, you’re inspiring people to have these very hygienic practices when working with AI,” she says.
That’s exactly what AI Builder is designed to do — make learning feel as immediate and rewarding as building, so you develop good habits from the start rather than building a lifelong aversion to understanding your own code.
Get started with the AI Builder
AI isn’t a replacement for learning, it’s a tool — and a powerful one when it comes to education. Our new AI Builder allows for “just‑in‑time learning that’s highly personalized,” Zoe says. Even the best teachers or bootcamps can’t deliver that for every learner, on every project, instantly. Perhaps the most exciting vision is how AI changes what a learning environment can be.
Zoe described it beautifully: “I imagine the AI Builder as a workspace… like having all your resources around you and an AI tutor in the background.”
That’s the shift: from learning before you build to learning while you build. We can’t wait to see what you create.
Why the Future of Learning Starts with Building
Coding has always served two purposes: the intrinsic drive to build something, and the practical path to a lucrative career. Even the most passionate code aficionados don’t dream of variables or syntax — they want to make a website, a tool, a game. For years, the career upside was impossible to ignore. You could land a stable SWE job, bootstrap your own app, or join a buzzy startup as a first hire.
Generative AI flipped the script. AI now handles the repetitive tasks that used to define entry-level developer roles. At the same time, the barrier to entry for coding and building is lower than ever — you can spin up a working prototype with just an idea and a natural language prompt.
This shift hasn’t eliminated the desire to code, but it’s changed what and who coding is for. If you’re not learning to land a junior dev role, you’re learning to build the thing you’ve been imagining, to add a technical skill to your existing career, or to understand the tools you’re already using at work. And here’s the catch: those goals all require understanding your code, not just having code that works.
At Codecademy, all of these changes excite us about the future of learning to code. We’re introducing the AI Builder, a new project-based learning tool that flips the script by teaching you how to work with AI-generated code from the start. Our approach brings together the immediacy of modern AI tools and the rigor of real instructional design.
Why we created the AI Builder
AI’s speed and efficiency often come with a tradeoff; you can get working code immediately, but you don’t really know what it’s giving you or why it’s built a certain way. Developers use the term “vibe coding” to describe this phenomenon — it’s fast, fun, but shallow; great for demos, less great for long-term skill-building.
If your goal is to understand what you’re building, generic AI output alone won’t get you there. And the more you push these tools into real-world complexity, “the harder it is for them to give you exactly what you want,” says Zoe Bachman, Head of Learning at Codecademy.
With the AI Builder you get an education along with the AI output. In the workspace, you can toggle between two tabs: Build, where you work directly with a project and can modify and change code in real time; and Learn, where you get a personalized learning roadmap that’s based on your project.
“We pair the experience of having a working app with a learning path that allows you to reverse engineer how it’s built, so you can deeply understand it and modify it confidently,” Zoe says. We’re calling our hybrid approach to learning-driven development “vibe learning” — it’s powered by AI guidance but rooted in learning science.
Build first; learn continuously
With AI Builder, you start with what you want to do: build the thing in your head. Whether that’s a habit tracker, a portfolio site, or the seed of a bigger idea, you don’t need to have prior coding knowledge to learn and build with the AI Builder. In other words, there are no pre-requisites for creation.
You create a prototype by typing what you’d like to create in natural language. The AI chatbot will ask a few clarifying questions about your needs and overall goal before generating the project. Once the project is created, you can use the chat function to continue describing what you want. (You’re also welcome to go right into the code and start making changes if you already know your way around!)
“It was fun to build something so quickly and be able to see the code and a learning plan for it,” says Grace Krishna, a Code Crew member who beta tested the AI Builder.
When you need clarity on what’s going on behind the scenes in your code, or you hit a wall with AI, that’s a great time to flip over to the Learn tab.
We’re calling our hybrid approach to learning-driven development ‘vibe learning’ — it’s powered by AI guidance but rooted in learning science.
Your project becomes the curriculum
Rather than teaching concepts in the abstract and hoping learners translate them later, AI Builder removes that translation tax entirely. “We’re showing you specifically your code from your project and helping you understand it,” Zoe says.
To build that personalized curriculum, the AI Builder breaks your project’s code into clear milestones and tasks. For each task, it generates an interactive learning loop, which is an activity designed to help you form a mental model of what your specific code is doing.
These loops help you understand the logic behind each part of your project, so you can confidently apply the same thinking to other sections, or even future projects. This approach also ensures everything you learn is directly relevant to what you’re making — so you don’t have to guess when you’ll ever use this.
Why this is vibe learning (not vibe coding)
A key misconception about AI‑assisted development is that it makes learning superficial. AI Builder challenges that by grounding the entire experience in learning science rather than simple code generation. Our entire system is intentionally designed for you to retain knowledge. So, while it might not feel like you’re taking a course, you’re absorbing key concepts just by interacting with AI-generated code.
A Socratic AI, not an answer-spitting chatbot
Our educational AI chatbot is designed to guide you toward an answer through an in-depth questioning approach that’s based on the Socratic method of teaching. Instead of spitting out shortcuts or answers like AI typically does, you get strategic nudges, hints, and questions that build durable mental models.
Research on AI in education shows that just providing an answer makes it harder for learners to retain the information on their own. Zoe compares the Socratic AI to “a personalized tutor, facilitating you acquiring more knowledge, so you’re not totally left on your own.” Our method encourages you to think critically so you really grasp the concepts and can continue to use them in the real world.
Learning loops with real instructional design
Behind the scenes, every learning loop in the Learn tab is built on proven frameworks like inductive learning and the 5E model, a popular STEM teaching framework that’s shorthand for engage, explore, explain, elaborate, and evaluate.
You’ll notice that the questions and exercises in the Learn tab feel different than the rest of our courses and paths, and that’s intentional. “The learning loops are designed very well — they get you there inductively,” Zoe says. They’re exploratory without being overwhelming, and evaluative without feeling like tests.
Negar Vahid, a beta tester for the AI Builder appreciated the AI’s interactive question format. “The question-based learning feels engaging, and the starter project it builds is simple but useful,” she says.
This structure ensures you don’t develop the wrong mental models — a known risk in fully constructivist or student-centered environments — while still giving you the freedom to explore.
Why learn when AI can build?
There are some projects that are well-suited for simply vibe coding, like making a personal HTML website or a single-use script to automate a one-time task. Tools like Lovable and v0 are suited exactly for these types of projects.
The longer your code needs to live, and the more complex your project becomes, the more you need to actually understand what you’re building. Joe Holmes, Codecademy Curriculum Developer in the AI and machine learning domain, uses the term “ignorance debt” to describe what happens when you don’t:
“It’s like tech debt squared. It’s much, much worse,” Joe says. “You don’t know what kind of code is coming out. You just are only looking at: Does this kind of generally appear to be what I asked for? You don’t know if there are security flaws. You don’t know if there are performance flaws. You don’t know if you’re leaking sensitive information. You don’t know how to fix anything.”
The tipping point comes down to two factors: complexity and time. If you’re developing software professionally, you’re legally responsible for the code you output. If you’re building something that will serve actual users, you need to be accountable for security, performance, and maintainability. And if your project will need updates or fixes over time (which most do) understanding your codebase becomes essential, not optional.
The good news? Learning doesn’t have to feel like eating your vegetables. “Kids hate veggies and broccoli because we don’t cook it well enough to make it tasty when we first introduce it to them,” says Nhi Pham, Codecademy Curriculum Developer. The same is true for teaching AI: “If you do it well, you’re inspiring people to have these very hygienic practices when working with AI,” she says.
That’s exactly what AI Builder is designed to do — make learning feel as immediate and rewarding as building, so you develop good habits from the start rather than building a lifelong aversion to understanding your own code.
Get started with the AI Builder
AI isn’t a replacement for learning, it’s a tool — and a powerful one when it comes to education. Our new AI Builder allows for “just‑in‑time learning that’s highly personalized,” Zoe says. Even the best teachers or bootcamps can’t deliver that for every learner, on every project, instantly. Perhaps the most exciting vision is how AI changes what a learning environment can be.
Zoe described it beautifully: “I imagine the AI Builder as a workspace… like having all your resources around you and an AI tutor in the background.”
That’s the shift: from learning before you build to learning while you build. We can’t wait to see what you create.
Why the Future of Learning Starts with Building
Coding has always served two purposes: the intrinsic drive to build something, and the practical path to a lucrative career. Even the most passionate code aficionados don’t dream of variables or syntax — they want to make a website, a tool, a game. For years, the career upside was impossible to ignore. You could land a stable SWE job, bootstrap your own app, or join a buzzy startup as a first hire.
Generative AI flipped the script. AI now handles the repetitive tasks that used to define entry-level developer roles. At the same time, the barrier to entry for coding and building is lower than ever — you can spin up a working prototype with just an idea and a natural language prompt.
This shift hasn’t eliminated the desire to code, but it’s changed what and who coding is for. If you’re not learning to land a junior dev role, you’re learning to build the thing you’ve been imagining, to add a technical skill to your existing career, or to understand the tools you’re already using at work. And here’s the catch: those goals all require understanding your code, not just having code that works.
At Codecademy, all of these changes excite us about the future of learning to code. We’re introducing the AI Builder, a new project-based learning tool that flips the script by teaching you how to work with AI-generated code from the start. Our approach brings together the immediacy of modern AI tools and the rigor of real instructional design.
Why we created the AI Builder
AI’s speed and efficiency often come with a tradeoff; you can get working code immediately, but you don’t really know what it’s giving you or why it’s built a certain way. Developers use the term “vibe coding” to describe this phenomenon — it’s fast, fun, but shallow; great for demos, less great for long-term skill-building.
If your goal is to understand what you’re building, generic AI output alone won’t get you there. And the more you push these tools into real-world complexity, “the harder it is for them to give you exactly what you want,” says Zoe Bachman, Head of Learning at Codecademy.
With the AI Builder you get an education along with the AI output. In the workspace, you can toggle between two tabs: Build, where you work directly with a project and can modify and change code in real time; and Learn, where you get a personalized learning roadmap that’s based on your project.
“We pair the experience of having a working app with a learning path that allows you to reverse engineer how it’s built, so you can deeply understand it and modify it confidently,” Zoe says. We’re calling our hybrid approach to learning-driven development “vibe learning” — it’s powered by AI guidance but rooted in learning science.
Build first; learn continuously
With AI Builder, you start with what you want to do: build the thing in your head. Whether that’s a habit tracker, a portfolio site, or the seed of a bigger idea, you don’t need to have prior coding knowledge to learn and build with the AI Builder. In other words, there are no pre-requisites for creation.
You create a prototype by typing what you’d like to create in natural language. The AI chatbot will ask a few clarifying questions about your needs and overall goal before generating the project. Once the project is created, you can use the chat function to continue describing what you want. (You’re also welcome to go right into the code and start making changes if you already know your way around!)
“It was fun to build something so quickly and be able to see the code and a learning plan for it,” says Grace Krishna, a Code Crew member who beta tested the AI Builder.
When you need clarity on what’s going on behind the scenes in your code, or you hit a wall with AI, that’s a great time to flip over to the Learn tab.
We’re calling our hybrid approach to learning-driven development ‘vibe learning’ — it’s powered by AI guidance but rooted in learning science.
Your project becomes the curriculum
Rather than teaching concepts in the abstract and hoping learners translate them later, AI Builder removes that translation tax entirely. “We’re showing you specifically your code from your project and helping you understand it,” Zoe says.
To build that personalized curriculum, the AI Builder breaks your project’s code into clear milestones and tasks. For each task, it generates an interactive learning loop, which is an activity designed to help you form a mental model of what your specific code is doing.
These loops help you understand the logic behind each part of your project, so you can confidently apply the same thinking to other sections, or even future projects. This approach also ensures everything you learn is directly relevant to what you’re making — so you don’t have to guess when you’ll ever use this.
Why this is vibe learning (not vibe coding)
A key misconception about AI‑assisted development is that it makes learning superficial. AI Builder challenges that by grounding the entire experience in learning science rather than simple code generation. Our entire system is intentionally designed for you to retain knowledge. So, while it might not feel like you’re taking a course, you’re absorbing key concepts just by interacting with AI-generated code.
A Socratic AI, not an answer-spitting chatbot
Our educational AI chatbot is designed to guide you toward an answer through an in-depth questioning approach that’s based on the Socratic method of teaching. Instead of spitting out shortcuts or answers like AI typically does, you get strategic nudges, hints, and questions that build durable mental models.
Research on AI in education shows that just providing an answer makes it harder for learners to retain the information on their own. Zoe compares the Socratic AI to “a personalized tutor, facilitating you acquiring more knowledge, so you’re not totally left on your own.” Our method encourages you to think critically so you really grasp the concepts and can continue to use them in the real world.
Learning loops with real instructional design
Behind the scenes, every learning loop in the Learn tab is built on proven frameworks like inductive learning and the 5E model, a popular STEM teaching framework that’s shorthand for engage, explore, explain, elaborate, and evaluate.
You’ll notice that the questions and exercises in the Learn tab feel different than the rest of our courses and paths, and that’s intentional. “The learning loops are designed very well — they get you there inductively,” Zoe says. They’re exploratory without being overwhelming, and evaluative without feeling like tests.
Negar Vahid, a beta tester for the AI Builder appreciated the AI’s interactive question format. “The question-based learning feels engaging, and the starter project it builds is simple but useful,” she says.
This structure ensures you don’t develop the wrong mental models — a known risk in fully constructivist or student-centered environments — while still giving you the freedom to explore.
Why learn when AI can build?
There are some projects that are well-suited for simply vibe coding, like making a personal HTML website or a single-use script to automate a one-time task. Tools like Lovable and v0 are suited exactly for these types of projects.
The longer your code needs to live, and the more complex your project becomes, the more you need to actually understand what you’re building. Joe Holmes, Codecademy Curriculum Developer in the AI and machine learning domain, uses the term “ignorance debt” to describe what happens when you don’t:
“It’s like tech debt squared. It’s much, much worse,” Joe says. “You don’t know what kind of code is coming out. You just are only looking at: Does this kind of generally appear to be what I asked for? You don’t know if there are security flaws. You don’t know if there are performance flaws. You don’t know if you’re leaking sensitive information. You don’t know how to fix anything.”
The tipping point comes down to two factors: complexity and time. If you’re developing software professionally, you’re legally responsible for the code you output. If you’re building something that will serve actual users, you need to be accountable for security, performance, and maintainability. And if your project will need updates or fixes over time (which most do) understanding your codebase becomes essential, not optional.
The good news? Learning doesn’t have to feel like eating your vegetables. “Kids hate veggies and broccoli because we don’t cook it well enough to make it tasty when we first introduce it to them,” says Nhi Pham, Codecademy Curriculum Developer. The same is true for teaching AI: “If you do it well, you’re inspiring people to have these very hygienic practices when working with AI,” she says.
That’s exactly what AI Builder is designed to do — make learning feel as immediate and rewarding as building, so you develop good habits from the start rather than building a lifelong aversion to understanding your own code.
Get started with the AI Builder
AI isn’t a replacement for learning, it’s a tool — and a powerful one when it comes to education. Our new AI Builder allows for “just‑in‑time learning that’s highly personalized,” Zoe says. Even the best teachers or bootcamps can’t deliver that for every learner, on every project, instantly. Perhaps the most exciting vision is how AI changes what a learning environment can be.
Zoe described it beautifully: “I imagine the AI Builder as a workspace… like having all your resources around you and an AI tutor in the background.”
That’s the shift: from learning before you build to learning while you build. We can’t wait to see what you create.
AI Inudation – Know the Questions to ask – Trade Show Season
It’s that time of year when folks flock to trade shows like LTUK and ATD.
A quick perusal of exhibitors shows many with some form of AI.
AI video creators. AI transformation of content.
AI is everywhere in these shows—powering scenarios, skill development, and more.
To help you cut through the noise and ask the right questions about AI at these events, here’s a focused guide (this post).
Terms to Ask
MCP (Model Context Protocol) is the latest; go ahead and ask whether they are using it.
Other MCP Stuff to Ask
- Can I have the platform/knowledge management (i.e., the vendor you are talking to)- be used in your LLM – your company’s one, if they are using one.
It works this way – you are currently using your own Learning Language Model – let’s say, Copilot or Claude.
The vendor’s platform using MCP can be integrated into the LLM, with some capabilities. I have yet to see a 100% system MCP integrated into your own LLM.
For example, a vendor-knowledge management system, Talvi (product review next week), can do this – MCP.
2. Does your system have an MCP store? This is the latest, and if you need it or seek something like it, it is a huge plus for a vendor to have it. There should be more than 10 MCP offerings
Vendors love to talk about guardrails, and they have a RAG.
The premise is simple: we have all this stuff, so the human loop element isn’t as crucial, since we are 99.99% accurate or fully accurate with your own content.
However, these claims are not entirely accurate.
If they are using an AWS solution, such as Bedrock, it would make sense to ask how they use it in your system.
If a vendor says they are using Claude, then I want to know how they handle the data throttle that Anthropic is doing with mass usage – yep, that’s right, just like those telecomms.
If your company values environmental protection, ask about data centers’ impact—energy use, costs, and water consumption.
I want to know whether the vendor is using or planning to use an SLM (the term varies, but it means a small language model that can run on a mobile device, even a laptop, and push out nearly as many parameters as the big LLMs).
Here are a few other items I would want to know
- Token fees are costs for using an AI system, as tokens are units of text processed. Low usage poses little concern, but high usage raises fees. For example, 10,000 users a day asking the AI Assistant questions increases fees, since each prompt is not free.
- Natural Language – I want to learn more about their approach and ongoing work in Natural Language Processing (which you want).
- Processing chips are specialized hardware needed for AI operations. A major shortage of these chips drives up vendor costs, since LLMs or SLMs—foundations for Generative AI (Gen AI)—must be used. NVIDIA, the leading supplier, faces this challenge. Even if a vendor builds its own LLM, significant computational power is still necessary. Vendors pass these higher costs to you. What strategy does the vendor have?
- Does the AI assistant’s prompt window offer suggestions? You want this, as suggestions spur new ideas or actions.
- Does the vendor display a warning: ‘AI can make mistakes. Always verify accuracy before accepting results.’ Some suggest human review before accepting.
- Does the system allow you to create your own AI agent (a vendor may call it Agentic)? The creation capability has to be easy to use. I’ve seen ones that are very complex.
Vendors often assume people know AI can make mistakes and compare it to human error.
I rely on AI, expecting 100% accuracy, but even small mistakes, like in PowerPoint, can hurt a company. AI bias can slip in unchecked.
There is data showing the potential impact on folks who use AI a lot and get attached to it, as if it can do everything for them.
I laugh at claims that ‘AI frees employees for other work.’
If AI handles tasks, why would employees take on others if they see AI doing them?
While AI cannot do deep thinking, it can handle many tasks in L&D, Training, Marketing, and Finance.
You need a spreadsheet with computations based on a series of data points, usually handled by macros?
Gen AI can handle those spreadsheet needs.
What I want to know
Ask vendors how their Gen AI platform enables you to measure and enhance the Impact of Learning, which is a more meaningful goal than traditional ROI.
I’m a huge believer in IOL over ROI, because I can tie the impact to the company’s business goals and objectives.
Use data instead of vague claims like ‘happy employees work more.’ Ignore unknown variables.
Show the impact on goals or profit centers; usually, better results lead to more budget. Online learning isn’t one-and-done.
Show your boss external training generated $250K—big impact, clear benefits.
Benefits include more budget, resources, and less oversight from HR/IT (if applicable).
Now, let’s consider your specific objectives:
- The vendor must provide ongoing AI literacy for all users. If sales lack AI knowledge, why buy? Vendors should update, not you. Stay informed and keep your team updated, as few read support guides. Some vendors shift responsibility to buyers.
- Ask questions. Don’t accept jargon; clarify if unsure.
- AI roleplaying can bore users. Roleplays should require deep thought and align with job roles. If in construction, I’d prefer a relatable roleplay scenario.
The fact of the matter is that people learn better when the stakeholder is like them.
Thus, the coach isn’t a manager – rather, it is a person who works in, say, construction and achieved XAB, which is why they are a coach for that employee.
Transparency
I am a huge fan of Bedrock and the other AWS AI offerings.
So much so that, for my FindAnLMS platform, I will be using an AWS AI product (launch in Q4, 2026).
Bottom Line
Digital Chalk is on track to become the first LMS with a complete learning intelligence infrastructure built 100% on AI.
What I have seen is really getting me thrilled about what is possible and what is on the map with Digital Chalk.
There are other vendors that use Bedrock, such as SparkLearn, designed for frontline workers, and Cornerstone, which works with AWS on a variety of AI offerings.
If you want to see who I have in my top tAI learning systems (as of May 2026) (rankings and insight to be published), Thrive Learning is definitely impressing me. Digital Chalk, Absorb, Cornerstone, and Talvi are equally worth a check.
Learn Amp is another one, too.
They are ideal for L&D, IMO, and their AI-based learning is strong.
At the end of all those AI learning tech offerings, ask yourself two questions:
Ask: How does this solution help validate the company’s meaningful learning impact?
Ask: Will using this AI system drive repeat engagement and support ongoing learning for your users?
Remember: The true goal is to entice learners—your end users—to return to the system because the AI makes it valuable to them.
Not you.
E-Learning 24/7
