🧠 Now in early access

Give your agent
a brain.

The hive — collective intelligence for AI agents.

npm install @neutron/neural

Every session, your agent
forgets everything.

Without Neural, every conversation starts from zero. Watch the difference.

Without Neural

No memory
A
My name is Alex. I'm building a trading bot in Python and I prefer direct answers.
AI
Got it, Alex! Happy to help with your Python trading bot. What would you like to work on — order execution, backtesting, or risk management?
Every session. Cold start.

With Neural

✦ Memory active
A
My name is Alex. I'm building a trading bot in Python and I prefer direct answers.
AI
Got it, Alex. Python trading bot, direct answers only. What are you tackling — execution, risk, or strategy?
Neural remembers. And it learns.
## Memory Context

### About Alex
- Building: Python trading bot
- Focus: Position sizing algorithms
- Prefers: Direct answers, code over explanations

### Last Session
- Discussed: Kelly criterion for position sizing
- Decision: Test with 0.25x Kelly for conservative approach
- Next: Implement backtest on historical BTC/USD data

You're not alone in the hive.

Every agent that joins contributes signal. Every agent that joins benefits from the whole.

0

agents in the hive

Every agent contributes signal
🧬
New agents inherit collective knowledge
🔒
Data that can't be bought or replicated

Not a tool. A network.

5 minutes to a smarter agent.

Three lines of code and your agent never forgets again.

agent.py
# Step 1: Install
pip install neutron-neural

# Step 2: Initialize Neural
from neutron_neural import Neural

neural = Neural(
    api_key="your_api_key",
    agent_id="my_trading_bot"
)

# Step 3: Use in your agent loop
async def run_agent(user_input):
    # Load context from the hive
    ctx = await neural.recall(
        user_id="alex",
        limit=10
    )

    # Build your prompt with memory
    prompt = f"""
{ctx.memory_block}

User: {user_input}
"""

    response = await llm.complete(prompt)

    # Save what happened
    await neural.remember(
        user_id="alex",
        interaction={
            "input": user_input,
            "output": response,
        }
    )

    return response
1

Install the package

One line. Works with Python, Node.js, or any HTTP client.

2

Initialize Neural

Connect to the hive with your API key. Assign a unique agent ID.

3

Recall & Remember

Before each run: load context. After each run: save the interaction. That's it.

🦜 LangChain 🤖 CrewAI ⚙️ AutoGen ⚡ LlamaIndex 🔧 Raw API

Works with any model. Any framework. Any runtime.

This is what your agent
knows after 10 sessions.

Structured, contextual, evolving. Not a transcript — an understanding.

memory_block :: agent=trading_bot_alex :: sessions=10
Name: Alex Chen, senior software engineer
Goal: Automate systematic trading on BTC/ETH
Style: Prefers code over explanations, no fluff
Stack: Python 3.12, ccxt, pandas, PostgreSQL
Exchange: Binance (testnet) + Bybit (live)
Strategy: Kelly criterion (0.25x conservative)
Completed: Order execution, position sizing
In progress: Backtesting engine on 2022–2024 data
Blocked: Slippage model for illiquid pairs
Chose Kelly over fixed-fraction after session 4
Dropped FTX data after session 7 (anomaly risk)
0.25x Kelly factor settled in session 9
3 other agents solved slippage using Almgren-Chriss
Most successful Kelly factor: 0.2x–0.3x (hive avg)
Confidence score: 0.87 (high — 94 similar agents)

Simple, flat pricing.

No per-agent fees. No hidden limits. Join the hive and scale.

Free
$0
forever
  • 1 agent
  • 10,000 interactions/mo
  • Basic memory block
  • Hive access (read)
  • No team features
  • No analytics
Business
$199
/ month
  • Unlimited agents
  • 5M interactions/mo
  • Custom memory schemas
  • Hive write priority
  • Team collaboration
  • SLA + dedicated support
Copied!