From Vibe Coding to Orchestrator: The 10 AI Terms That Redefined What Developers Do In 2025

I want to take this chance at the end of the year to reflect on these new terms, what they mean and how they have changed the landscape of software.

I personally think number 9 is and will continue to be the most popular.

Disclaimer: This talk: https://www.youtube.com/watch?v=FoXHScf1mjA from Addy Osmani was the main inspiration for this post.

1. Vibe Coding

This term was brought into being by Andrej Karpathy. He definitely tapped into something because the usage exploded this year. Vibe Coding is where the user creates code purely by communicating with an AI chatbot via prompts.

Over the year the term has become synonymous with “poor quality code that does the job” or “some code put together quickly for an MVP”. It’s usage is for rapidly creating some output where speed is prioritised over quality.

Impact: Many people who previously were on the periphery of software are now enabled to build out their ideas themselves. It’s much easier and quicker to build out MVPs. The production of code is now very cheap so much more code is being written. However, more code doesn’t mean better software.

2. AI Agent

This refers to a software system that can autonomously perform coding tasks with minimal human supervision. The user would provide a prompt — instructions in a written language — and the AI agent would try to complete those instructions generally unassisted.

An example would be where you ask an AI agent in the Cursor Editor to code out a feature on a Web application — like a new Navigation bar.

Impact: 90% of developers are using AI in their daily work now and the majority are using some kind of AI agent to interact with. This is a huge change in workflows and DX. The rapid adoption of tools like Cursor just go to show how bad the DX was previously for most developers. Oh and StackOverflow is a thing of the past.

3. Context Window

The “Context Window” is the maximum amount of data or text an AI can work with at one time. It acts as the AI’s working memory so to speak. The context an AI has to work with is a critical element in the experience the user has when interacting with an AI model.

The “Context Window” is always a point of discussion when dealing with AI. Was the Context Window too big and the AI got confused or was the Context Window too small and the AI didn’t have enough context to produce a good answer.

Impact: If you are working with AI then you have to be aware of context and the context window.

4. Memory Bank

This is a structured documentation system. This allows the user to retain context over multiple sessions by storing certain pertinent information in files (markdown) which act as a so called “memory”. The AI agent can then refer to these documents when needed in order to gather the right context.

Impact: Memory does play an important role in the usefulness of AI in your day to day work. If you have a specific way to do things that you want AI to repeat regularly then it’s “memory bank” can be very useful to help with this.

5. Prompt Engineering

This is where the user would craft “prompts” in just the right way in order to get the AI to output reliable and repeatable code. There have been many attempts made at getting the “perfect prompt” but it’s rarely worked over any length of time.

Impact: When this term first came out it seemed we would all need to learn some “magic formula” in order to manipulate the AI in just the right way to give us the output that we want. However, turns out developers don’t like this to do this and other approaches have been adopted.

6. Spec Driven Development

This is where the user would write out a very detailed plan before engaging with the AI. The hope is that if you give such specific instructions then the AI would have less chance to produce output the user didn’t want or ask for. Some developers love this approach, others really don’t.

Impact: The theory makes sense but at what point would it not just be better to code out the thing yourself? The main issue though is the results given here don’t seem to be much better than what you would get with much shorter prompts. Is it worth all that upfront effort? Knowing developers, probably not.

7. Background Agents

These are AI agents that run in the background as the developer focuses on other tasks. This already exists in tools like Cursor or Jules from Google labs. The developer can prompt the background agent to do some task and then let it go off and do its thing. Then the developer can come back later and check the output.

Technically the developer can be anywhere when using these agents. You could be going into the cinema, ask the agent to do a task, and when the film is finished the task will be done. That’s the theory anyway.

Impact: This has the potential to have a huge impact if the DX can be achieved. The main problems at the moment are:

• Having the mental capacity to manage all these background agents.
• Cost — Having many background agents running can get expensive.
• Maintaining consistency — It’s hard to maintain quality when having so many agents to manage.
• You turn into a manager rather than a developer. Some people don’t want that.

8. MCP

The Model Context Protocol. A lovely fancy term but is it here to stay? It’s an open standard for connecting LLMs to data and tools. For example you have the Atlassian MCP which can be used to create Jira tickets or Confluence documents. The theory is beautiful, the reality is different.

Impact: It seems that developers aren’t using MCPs that much. This still isn’t clear and it depends on who you listen to. Will they be around in a year? Who knows. It’s good to know about them but don’t commit too hard to them just yet.

9. AI-Assisted Engineering

This is where a developer will be developing along side an AI agent but keep a tight leash.

The main difference here from Vibe Coding is the human remains in control and takes full responsibility for the output. This is working the best in established code bases with experienced developers.

Impact: This currently seems to be the most popular way to using AI in your current workflow.

This has had a massive impact on the industry with reliable reports stating 30% efficiency gains on a daily basis. The effect this will have on Junior developers over the long run is still unclear.

10. Orchestrator Of Coding Agents

Developers who are more like “managers” that delegate all the implementation tasks out to AI coding agents who produce the actual code. The developer would then have a constant high level view of the software and thus be able to guide the AI agents to carry out needed tasks.

Impact: This hasn’t really caught on yet in software. The tools are all there but the average engineer is not really doing this. Why not? I think the main reason is it’s just a totally different approach to what a developer traditionally does. This is more like managing than developing and most developers don’t want to be doing this. Will this change over time? Maybe as a younger generation come in but only time will tell.

Conclusion

As you can see, there are many news terms that have entered the software world in the past 12 months. AI is not replacing developers but it is changing the way they work.

Whether you like this or not, I think it’s important to be aware of the terminology and where the industry is heading. You don’t have to follow it religiously and jump on every new hype train but you probably should be keeping an eye on it. I don’t see 2026 being very different to 2025 in this regard.

What has your experience been in 2025? What do you think the biggest impact items were? Did I miss anything obvious off this list?

Subscribe to My Weekly Updates on Medium!

Enjoyed This Post?

If you found this blog post helpful, why not stay updated with my latest content? Subscribe to receive email notifications every time I publish.

If you're feeling really generous you can buy me a coffee. (Btw, I really like coffee…)

What You'll Get

• Exciting Discoveries: Be the first to know about the latest tools and libraries.
• How-To Guides: Step-by-step articles to enhance your development skills.
• Opinion Pieces: Thought-provoking insights into the world of frontend development.

Join Our Community

I live in the vibrant city of Prague, Czech Republic, with my family. My blog is more than just articles; it's a community of like-minded developers who share a love for innovation and learning.

About me

I'm a passionate Frontend Developer specialising in React and TypeScript. My professional journey revolves around exploring and mastering new tools and libraries within the JavaScript ecosystem.

Check out my LinkedIn and Github if you are interested.