AI Radar
What is buzzing in AI
New words, tools and trends, explained in plain English, so you always know what people are talking about.
Updated 6 hours ago10 new this weekGet it by emailRSS feed
BuzzwordRisingTop storyModel Context Protocol (MCP)
MCP is an open standard that lets AI apps connect to tools and data in one common way, so nobody has to build a custom link for every app and tool pair.
ReadClose
Model Context Protocol (MCP)
MCP is an open standard that lets AI apps connect to tools and data in one common way, so nobody has to build a custom link for every app and tool pair.
Why it matters
Job posts for agent roles now list MCP. It is the quickest way to give an assistant safe, controlled access to your files, tickets and databases.
BuzzwordRisingAgentic AI
Agentic AI plans and takes steps on its own, using tools, instead of only answering one question at a time. It can search, decide, act and check its work.
ReadClose
Agentic AI
Agentic AI plans and takes steps on its own, using tools, instead of only answering one question at a time. It can search, decide, act and check its work.
Why it matters
Most new AI jobs, from agent engineer to product manager, are about building, running or judging these systems.
BuzzwordRisingContext engineering
The skill of choosing what a model sees for a task: the instructions, the right documents, the memory and the tool results, and leaving out the rest.
ReadClose
Context engineering
The skill of choosing what a model sees for a task: the instructions, the right documents, the memory and the tool results, and leaving out the rest.
Why it matters
Better context beats a cleverer prompt. It is why two teams using the same model get very different results.
TrendRisingVoice agents
AI that holds a spoken conversation: it listens, thinks and answers in a natural voice, so it can book, answer or support callers on the phone.
ReadClose
Voice agents
AI that holds a spoken conversation: it listens, thinks and answers in a natural voice, so it can book, answer or support callers on the phone.
Why it matters
Businesses are adding them to phone lines, and people who can build and test one are in demand.
Want to build one? Follow the voice agent playbook, which takes you from a phone call to a safe, tested agent in eight steps.
BuzzwordSteadyVibe coding
Describing what you want in plain language and letting an AI write the code, then guiding it by testing the result instead of writing every line yourself.
ReadClose
Vibe coding
Describing what you want in plain language and letting an AI write the code, then guiding it by testing the result instead of writing every line yourself.
Why it matters
It lets non-programmers build prototypes. You still have to test, read and secure what it makes.
ExplainerRisingAI evals
Evals are repeatable tests that score an AI feature on real cases, so you can tell whether a change made it better or worse.
ReadClose
AI evals
Evals are repeatable tests that score an AI feature on real cases, so you can tell whether a change made it better or worse.
Why it matters
Teams that ship reliable AI test it on every change. Evaluation is now a job title of its own.
BuzzwordSteadyReasoning (thinking) models
Models that spend extra time working through a problem step by step before they answer. That helps with hard maths, code and planning, at higher cost and delay.
ReadClose
Reasoning (thinking) models
Models that spend extra time working through a problem step by step before they answer. That helps with hard maths, code and planning, at higher cost and delay.
Why it matters
Knowing when to turn thinking on, and when it is wasted money, is a practical skill for any AI builder.
TrendRisingSmall language models
Compact models that run on a laptop, a phone or a small server. They are cheaper and faster, keep data on your device, and can be very good at one focused job.
ReadClose
Small language models
Compact models that run on a laptop, a phone or a small server. They are cheaper and faster, keep data on your device, and can be very good at one focused job.
Why it matters
They make private and offline AI possible, and they cut the cost of high-volume tasks.
ExplainerSteadyPrompt injection
A trick where hidden instructions in a web page, email or file make an AI follow an attacker's orders. It is the main security risk for agents that read outside content.
ReadClose
Prompt injection
A trick where hidden instructions in a web page, email or file make an AI follow an attacker's orders. It is the main security risk for agents that read outside content.
Why it matters
Anyone building agents is expected to know how to limit the damage a fooled model can do.
BuzzwordNewAgent-to-agent (A2A)
A young standard that lets AI agents from different makers hand tasks to each other and follow progress, the way colleagues would.
ReadClose
Agent-to-agent (A2A)
A young standard that lets AI agents from different makers hand tasks to each other and follow progress, the way colleagues would.
Why it matters
It is how separate agents may one day cooperate across companies. It is still early, so expect it to change.
Free · weekly
Get the AI Radar in your inbox
The new buzzwords and trends of the week, in plain English, and why they matter. No spam; unsubscribe any time.