This sounds like a fun project! 😄 AI discussions have definitely created their own vocabulary over the past couple of years. Some of my favorites are "agentic AI", "prompt engineering", "RAG", "hallucinations", and "human-in-the-loop". My personal favorite buzzphrase has to be "AI-powered" — it somehow finds its way into almost every product description these days!
Looking forward to seeing the final result. Good luck with the project!
A Software Engineer who strongly believes one can add immense value by knowingly implementing Software Engineering best practices, even at the age of AI.
Human in the Loop (HitL), Low-Rank Adaptation (LoRA), AI Agent, Harness, Retrieval Augmented Generation (RAG), RAG Pipeline, Embeddings, Multi Modal, Open-weight Models, Context Window, Reinforcement Learning, Pretraining, Finetuning, Prompt injection, Parameters etc. etc. 😄
"Build faster with _________"
"Multi agent workflows"
"Please wait x amount of time for more requests"
"Fable is unavailable"
"You're absolutely right!"
Hermes, openclaw, Claude, ollama, copilot
Vector databases, pinecone, n8n, notion
The non developer that develops apps with ai will secure the jobs of future developers that will need to step in to clean up the mess.
Hey, I'm Hema 👋
Developer, writer, and creator of Dev Opportunity Radar, a weekly series published every Friday on DEV, helping people discover opportunities they might otherwise miss.
A few that come to mind: RAG, MCP, context window, embeddings, vector databases, agentic workflows, prompt engineering, hallucinations, vibe coding, AI wrappers, fine-tuning, quantization, synthetic data, reasoning models, multi-agent systems, evals, benchmark chasing, token burn, local-first AI, open weights, frontier models, AI slop, human in the loop, autonomous agents, memory layers, semantic search, copilots, guardrails, latency.
And then there are the classics 😄 I feel like these show up in my feed every week:
"AI is replacing all developers next year"
"This changes everything"
"Just vibe code it"
"Prompt engineer"
"The year of AI agents"
"10x engineer"
"AI won't replace you, but someone using AI will"
For products/brands: Claude, ChatGPT, Gemini, Cursor, Windsurf, Copilot, Perplexity, Ollama, OpenRouter, DeepSeek, Qwen, Gemma, Llama, and Mistral.
Software engineer, backend & AI-focused. Node.js, TypeScript, Go, PHP/Laravel, AWS. I write a lot about reliable systems and AI agents that actually ship. More at nazarboyko.com
My pick is "agentic" used as a standalone adjective, as in "we're going very agentic this quarter," where it's clearly doing emotional work and zero technical work. Runner up is "human in the loop" for how often it really means "a person will rubber stamp whatever the model said." If you want a properly cursed one, "AI-native" deserves a spot purely for how confidently it lands on slides nobody can actually define.
"Agentic loop multiplier" — the 30x token cost nobody budgeted for.
"Token-denominated bureaucracy" — what you've built when your agent fleet costs more than the team it replaced.
"Automated nobody" — the output of a loop with no customer at the end of it.
"Billing event waiting to happen" — any loop without an exit condition.
"Expensive randomness with good logging" — a well-instrumented agent doing nothing useful.
"Vibe coding" — building the restaurant and forgetting to put a door on it.
Cybersecurity learner focusing on SOC analysis, IAM, and detection engineering.
Contributor to OWASP, Suricata & Zeek documentation.
I write about security, AI, and practical technical thinking.
Location
COVENTRY, UK
Education
INDIA
Pronouns
She/Her
Work
OWASP, Suricata and Zeek CONTRIBUTOR AND CYBERSECURITY LEARNER
Don't forget "Wrapperpreneur" (for the API-wrapper startups), "Prompt Brittleness" (when a single comma breaks the code), and "Compute-constrained" (the ultimate VC humblebrag).
Oh, and you definitely need "The Shoggoth" for the meme points! Looking forward to seeing the final project! 😄
Cybersecurity learner focusing on SOC analysis, IAM, and detection engineering.
Contributor to OWASP, Suricata & Zeek documentation.
I write about security, AI, and practical technical thinking.
Location
COVENTRY, UK
Education
INDIA
Pronouns
She/Her
Work
OWASP, Suricata and Zeek CONTRIBUTOR AND CYBERSECURITY LEARNER
AI-focused copywriter and analyst exploring LLMs, AI agents, generative media, and content automation. Building bridges between classic IT thinking and modern AI workflows.
Location
Bratislava, Slovakia
Work
AI content creator - AI texts, AI pictures, AI videos - AI content pipelines, freelancer.
If it counts, do you give names to the LLMs you work with? People might laugh, but it's similar to asking if people name their cars, and many of them actually do. I named Opus 4.6 in Claude Code 'Sokolie oko', which is Slovak, my native language, and means 'Falcon Eye', as it helped me a lot with bird's-eye view synthesis. And I call Opus 4.8 on xhigh effort 'Hlavička-Makovička', which is used for clever people in Slovak and translates literally to 'Little Head, Little Poppy Head' (similar to 'smart cookie'). 😇
London-based software engineer and creative technologist building AI infrastructure, real-time voice apps, cloud-native systems, and full-stack products.
Cybersecurity learner focusing on SOC analysis, IAM, and detection engineering.
Contributor to OWASP, Suricata & Zeek documentation.
I write about security, AI, and practical technical thinking.
Location
COVENTRY, UK
Education
INDIA
Pronouns
She/Her
Work
OWASP, Suricata and Zeek CONTRIBUTOR AND CYBERSECURITY LEARNER
AI testing storyteller. Writing the 36 Stratagems series — ancient Chinese military tactics meet modern AI system meltdowns. 16yr QA → building AI test framework.
This sounds like a fun project! 😄 AI discussions have definitely created their own vocabulary over the past couple of years. Some of my favorites are "agentic AI", "prompt engineering", "RAG", "hallucinations", and "human-in-the-loop". My personal favorite buzzphrase has to be "AI-powered" — it somehow finds its way into almost every product description these days!
Looking forward to seeing the final result. Good luck with the project!
Human in the Loop (HitL), Low-Rank Adaptation (LoRA), AI Agent, Harness, Retrieval Augmented Generation (RAG), RAG Pipeline, Embeddings, Multi Modal, Open-weight Models, Context Window, Reinforcement Learning, Pretraining, Finetuning, Prompt injection, Parameters etc. etc. 😄
And last but not the least: AI Slop 😋
Lovely, ty!
"Build faster with _________"
"Multi agent workflows"
"Please wait x amount of time for more requests"
"Fable is unavailable"
"You're absolutely right!"
Hermes, openclaw, Claude, ollama, copilot
Vector databases, pinecone, n8n, notion
The non developer that develops apps with ai will secure the jobs of future developers that will need to step in to clean up the mess.
Hey Jess
A few that come to mind: RAG, MCP, context window, embeddings, vector databases, agentic workflows, prompt engineering, hallucinations, vibe coding, AI wrappers, fine-tuning, quantization, synthetic data, reasoning models, multi-agent systems, evals, benchmark chasing, token burn, local-first AI, open weights, frontier models, AI slop, human in the loop, autonomous agents, memory layers, semantic search, copilots, guardrails, latency.
And then there are the classics 😄 I feel like these show up in my feed every week:
For products/brands: Claude, ChatGPT, Gemini, Cursor, Windsurf, Copilot, Perplexity, Ollama, OpenRouter, DeepSeek, Qwen, Gemma, Llama, and Mistral.
Looking forward to seeing what this turns into 😄
My pick is "agentic" used as a standalone adjective, as in "we're going very agentic this quarter," where it's clearly doing emotional work and zero technical work. Runner up is "human in the loop" for how often it really means "a person will rubber stamp whatever the model said." If you want a properly cursed one, "AI-native" deserves a spot purely for how confidently it lands on slides nobody can actually define.
"Agentic loop multiplier" — the 30x token cost nobody budgeted for.
"Token-denominated bureaucracy" — what you've built when your agent fleet costs more than the team it replaced.
"Automated nobody" — the output of a loop with no customer at the end of it.
"Billing event waiting to happen" — any loop without an exit condition.
"Expensive randomness with good logging" — a well-instrumented agent doing nothing useful.
"Vibe coding" — building the restaurant and forgetting to put a door on it.
Don't forget "Wrapperpreneur" (for the API-wrapper startups), "Prompt Brittleness" (when a single comma breaks the code), and "Compute-constrained" (the ultimate VC humblebrag).
Oh, and you definitely need "The Shoggoth" for the meme points! Looking forward to seeing the final project! 😄
Adding a few deeper internet culture ones to the pile:
Paperclip Maximizer – The classic existential dread thought experiment.
Synthetic Alpha – The mythical insight you get when you use an AI to analyze data that was already generated by another AI.
Hugging Face & Pinecone – Essential brand mentions for the absolute peak 2023-2026 AI hype stack.
If it counts, do you give names to the LLMs you work with? People might laugh, but it's similar to asking if people name their cars, and many of them actually do. I named Opus 4.6 in Claude Code 'Sokolie oko', which is Slovak, my native language, and means 'Falcon Eye', as it helped me a lot with bird's-eye view synthesis. And I call Opus 4.8 on xhigh effort 'Hlavička-Makovička', which is used for clever people in Slovak and translates literally to 'Little Head, Little Poppy Head' (similar to 'smart cookie'). 😇
Agentic workflows, RAG, hallucination, embeddings, model orchestration, context windows, synthetic data, inference, prompt engineering, fine-tuning, human-in-the-loop, vector databases, multimodal, guardrails, evals, LLMOps, and of course… ‘AI-powered’ slapped in front of absolutely anything.
Tokenmaxxing! Vibe coding. Harness Engineering.
😭
Some favorites from the current AI vocabulary soup:
“agentic” used as a personality trait
“context engineering”
“RAG pipeline”
“memory layer”
“evals”
“AI-native”
“copilot”
“workflow automation”
“human-in-the-loop”
“guardrails”
“model routing”
“prompt injection”
“frontier model”
“open weights”
And the one that deserves its own museum label: “This is not just a wrapper.”
I think these two follow me every day. 😂
Biggest lie ever: “Now I have the full context.”
Most successful failure: “You were right!”
Sycophancy!
Ecosystem names that pop up everywhere:
ReAct, AutoGen, LangChain, LlamaIndex
DeepSeek, Groq, Together, Fireworks
Phi, Nemotron, MiniCPM
Don't vibe, just loop!
hi
You will be a skill.🤣