nero is basically an AI assistant that doesn't suck. i got tired of chatgpt being boring and having no personality so i built my own thing. it has 7 different brains that handle different types of tasks, it can actually speak back to you with natural voices, and the whole UI looks like something out of a sci-fi movie.
oh and it's completely free to use if you plug in a gemini api key (which is also free lol).
nero routes your messages to the best brain for the job:
| brain | what it does |
|---|---|
| reasoning | logic, analysis, step-by-step explanations |
| coding | writes code, fixes bugs, runs scripts |
| research | deep dives, web search, facts |
| creative | stories, brainstorming, writing |
| memory | remembers stuff about you, takes notes |
| learning | teaches concepts, tutorials, analogies |
| automation | scripts, workflows, scheduling |
yeah it literally talks. pick from 8 neural voices powered by gemini tts.
it remembers things. you tell it your name, your preferences, stuff you're working on — it stores all of that and brings it up later. there's emotional intelligence too, so it can tell when you're stressed and suggest you take a break.
it builds a knowledge graph from your conversations. topics you discuss get connected automatically, so it can show you how different things you've talked about relate to each other.
create projects, manage files, track tasks — all from inside the AI. it's not just a chatbot, it's more like a workspace.
- node.js 18+ (i'm using 20 but whatever works)
- a gemini api key — get one free at Google AI Studio
# clone it
git clone https://github.com/DeepMane33/nero-ai.git
cd nero-ai
# install deps
npm install
# add your api key
cp .env.example .env.local
# edit .env.local and add your GEMINI_API_KEY
# run it
npm run devthen open http://localhost:3000 and you're in.
GEMINI_API_KEY=your_key_herethat's literally all you need. the app works with just gemini. if you want more providers you can add groq, openrouter, cerebras, sambanova, or anthropic keys too — but gemini alone gets you pretty far.
- next.js 16 — app router, server components, turbopack
- react 19 — the latest
- typescript — because i'm not a maniac
- tailwind css 4 — styling
- better-sqlite3 — local database, no setup needed
- framer-motion — animations
- gemini tts — natural voice synthesis
- swr — data fetching
nero-ai/
├── src/
│ ├── app/ # next.js app router pages + API routes
│ │ ├── api/ # all the backend endpoints
│ │ │ ├── chat/ # main chat endpoint
│ │ │ ├── tts/ # gemini text-to-speech
│ │ │ ├── vision/ # image analysis
│ │ │ ├── search/ # web search
│ │ │ └── ...
│ │ └── page.tsx # dashboard
│ ├── components/ # react components
│ │ ├── chat/ # chat interface, messages
│ │ ├── voice/ # voice mode with waveform
│ │ ├── coding/ # code editor, terminal, file explorer
│ │ ├── ui/ # shared UI components
│ │ └── ...
│ ├── core/ # brain models, tools
│ ├── lib/ # utilities, database, APIs
│ └── contexts/ # theme context
├── soul.md # nero's personality definition
└── package.json
honestly? because i wanted an AI that felt like mine. chatgpt is great but it's generic. gemini is good but the interface is boring. i wanted something with:
- a personality that i control
- voice that actually sounds good
- tools that let it DO things, not just talk
- a UI that doesn't look like a corporate product demo
- memory so it actually knows who i am
and yeah it took a while to build but it works and i use it every day so...
if you want to help out, go for it. fork it, mess around, submit a PR. just don't break the voice mode because that took forever to get right.
mit.